Raid0 Btrfs grub boot experiments full size devices (See Edits)

lilsirecho · 2012-07-15 01:29:54

Latest upgrade running well:

sh-4.2# uname -a
Linux n6re 3.4.4-3-ARCH #1 SMP PREEMPT Tue Jul 3 14:36:44 UTC 2012 x86_64 GNU/Linux
sh-4.2#

DSpider · 2012-07-15 06:27:32

Speed has nothing to do with GRUB2. Seriously, nobody cares.

Mark this thread as solved and move on.

lilsirecho · 2012-07-16 00:46:11

Basic to this post is the fact that grub2 does not normally permit booting from a partition.

Booting from a partition with CF cards is as fast as is possible with no seek-time such as occurs with HDD's.

Seven seconds boot is obtained with 4gb of installed packages with grub2 partitioned boot.

Grub2 doesn't speed it up but does require attention to boot from a partition.

Basic with this posting is also that it is root only and raid0 to boot from and operate within....striped raid with three cf devices.

My experiment is not yet over, however. Still have raid within raid to investigate.

Appreciate your remark Dspider and herewith responded.

lilsirecho · 2012-08-24 03:24:17

Latest experiement was performed this date.

Installed raid) grub2 bootable into a single 8GB CF card using two partitions for the raid pair.

The CF card used is a ADATA 8GB 266x device. The speed of the device in this mode is not twice that of the basic CF. All elements report 45mb/s including the raid device.

I am posting this from the booted raid0 card..

Perhaps there is a benefit from this arrangement which uses two partitions for the raid0. At least the card stands alone in the raid system so a lost device can't destroy the raid(unless its the card itself!)

It boots in about ten seconds to xfce4.

So single device raid devices can be utilized but benefits are questionable.

It was fun doing it!!!!

lilsirecho · 2012-08-25 21:42:46

Built another raid0 partitioned grub2 bootable array with a single CFcard 32GB today.

Installed with a 60MB unassigned sector before the partition #1 to allow for grub MBR.

Assigned 60MB for boot on partition1 ext2.

Assigned half of remaining space for partition2 ext3 and the other half to partition3 ext3.

Installed per the outlined procedure posted previously.

Boots directly into xfce4 and includes firefox, feh, vlc, leafpad and jumanji at this time.

EDIT: Boots in 10 seconds.

Last edited by lilsirecho (2012-08-25 21:44:36)

lilsirecho · 2012-09-05 04:41:44

Remodeled the raid system to use a pcie to sataII device, permitting the array to be 4 CF cards.

The install fails to boot from default grub2 but boots fine from fallback. This indicates some missing kernel module(s).

The hdparm read speed obtains 280MB/s.

I have yet to try the dd method which is more indicative of performance which includes read and write results.

This 4 CF card array is still raid0 with a boot partition and rootpartition. It gives an array size of 59GB with four UDMA 16GB cf cards.

It boots in 13 seconds to xfce4 desktop. Includes jumanji browser, firefox,leafpad ,kdenlive, gparted,hdparm ,feh.

lilsirecho · 2012-09-08 21:47:10

Utilized the latest .iso and produced a single CF card raid0 with 3 partitions:x86_64 with the outlined steps shown.......................CF Card has free space of 60MB at beginning for grub data install.

Utilized pre-partitioned CF card and skipped cfdisk in the install sequence, going into:

mdadm --create /dev/md7 --level-0 --raid-devices=2 /dev/sda2 /dev/sda3...then:
mke2fs /dev/sda1 for /boot and generated ext3 on the /dev/md7.

Mounted /dev md7 /mnt.

Performed....pacstrap /mnt base base-devel, genfstab.
Entered: arch-chroot /mnt
Entered :mount /dev/sda1 /mnt/boot

Following that, configured the system including locale-gen. Also added mdadm hook in mkinicpio.conf. Steps not mentioned also performed.

Setup /etc rc.conf for static ip and /etc/resolv.conf for nameservers.

Setup /etc/pacman.conf for "Never" in all repos and default.

Verified ....pacman -Syy..... performed correctly.

Because the system is now in chroot, I can install grub2 and setup the raid0 to boot from partition /dev/sda1.

pacman -S grub2-bios
grub-mkconfig -o /boot/grub/grub.cfg
grub-install /dev/sda........  (mbr)
mdadm --examine --scan > /etc/mdadm.conf
mkinitcpio -p linux
This must provide an mdadm comment and not be missing bsdcpio (default locale)  If it is missing, redo locale.gen and repeat the mdadm step and mkinitcpio -p linux.

The install is now finished.......exit the chroot and reboot to the raid0 CF card.

Hope there are no oops!

Last edited by lilsirecho (2012-09-08 21:49:57)

lilsirecho · 2012-09-08 21:57:24

Added .....init=/bin/systemd to the linux line and booted into this forum for this post.

Boot time a bit long probably due to mdadm activity, somewhat more than 10 seconds.

lilsirecho · 2012-09-12 15:40:56

A new experiment using a single CF card and two partitions with no boot nor swap, all root.

Began with cfdisk of a data 8GB 266x CF card to delete all and write partition table. Reboot.

Enter gparted and select the drive. Install 60mb free space ahead of the two partitions which split the remaining capacity into equal ext3 partitions.

Using the arch 2012-08-18 .iso, entered ip link set eth0 data followed by /etc/resolv.conf to enter nameservers in the dhcp resolv file.

Entered alt-f2 and root login.

Performed:

#mdadm --create /dev/md9 --level=0 --raid-devices=2  /dev/sdb1 /dev/sdb2

This may produce a comment and ask ...Continue?...answer Yes.

The array is ready when the reply is .../dev/md9 has been started with 2 drives.

mke2fs  -j /dev/sdb1......(when finished,do:
mke2fs -j /dev/sdb2

When finished, exit with... alt f1 ...to return to archiso sequences.

Now set up /etc/rc.conf..../etc/pacman.conf for NEVER....verify resolv.conf.

Perform pacman -Syy and reply should provide repo sync...troubleshoot until it syncs with this command.

Perform :

#mount /dev/md9 /mnt
@pacstrap  /mnt base base-devel

The packages should now be downloaded and installed.

Following the package install, the genfstab step is performed.

One entry in /etc/mkinitcpio.conf hooks needs to be added....mdadm.

After doing the mirrorlist , recheck ... pacman -Syy (checks the status of internet)

The following sequences establish raid0 booting with mdadm.......

#pacman -S grub2-bios
#grub-mkconfig -o /boot/grub/grub.cfg
#grub-install /dev/sdb....................( installs MBR)
#mdadm --examine --scan > /etc/mdadm.conf....(this should return nothing)
#mkinitcpio -p linux
(This will perform the normal sequences except it adds a mdadm comment that mdadm.conf arrays will be used and this indicates that the array is established for booting /dev/md9.  Latest kernels utilize array ID's such as /md127 and the like and do often appear in examining array status .  A label assigned helps to identify the desired array among several)

#exit
#umount proc sys dev
#reboot

I post this from my single CF card /root mdadm array boot-up in xfce4. The boot time for this particular card is about 20 seconds because it isn't full UDMA capable. It is good enough to demonstrate a variation in raid0 using no boot partition nor swap partition

I cannot eliminate root! With the maxell CF cards having UDMA capability, I expect much faster boot. I have to find another as all are in use.

I hope to use this mode for fast boot into internet and e-mail. No terrabytes here!

lilsirecho · 2012-09-19 03:10:04

One last note:

My latest raid0 grub2 bootable partitioned system is now permanent systemd and boots into xfce4 Desktop in 12 seconds with kernel 3.4.5-1. It operates in root mode. I do not expect to run readahead on this small system which is 8GB Compact Flash 266x.

I would expect maxell UDMA 16GB CF card would boot some four seconds faster. It also provides more capacity!

EDIT: The speed expected did not occur. Have to look into that.

Have six such systemd raid0 systems, some with two separate devices, one HDD, and four single raid0 devices all bootable in grub2 partitioned boot. Other than the hdd the devices are CF cards.

EDIT: For some time I haven't been able to boot the raid arrays from USB. Today, Oct 28,2012 I have successfully done that! To perform the feat, in mkinitcpio.conf HOOKS add sleep after udev and place mdadm after sata and usb usbinput before filesystems followed by usr shutdown fsck. Have to do this in chroot and run mkinitcpio -p linux.

With my devices in systemd using this arrangement on single device raid0 setups, the boot time is quite a bit longer. There may be a better arrangement of HOOKS to permit removing the sleep entry.
EDIT: Recent installs of systemd raid0 via archiso for October utilized the arch-chroot /mnt to enable the grub-mdadm pair to be installed during the archiso install procedure rather than doing that on a separate chroot. I also plan to try other arrangements of HOOKS hopefully to permit removing the sleep hook. All devices are now with kernel 3.6.3-1.

EDIT: Point A: No luck with other methods to set up boot in usb.

Systems are now up in linux 3.6.6-1. It seems the boot time increases with upgrades.

Am wary of systemd parallel operations utilized with raid0 striping (also parallel in operation) which might cause the two to operate separately thus causing excessive boot time. Not aware of the exact performance of either in the boot sequence. Suspect that raid0 demands full control during the build.

Discovered that incorrect hdware time produced xfce4 Desktop but no keybd nor mouse. Correctling the time solved the discrepancy.

Would appreciate any comments regarding the boot time within systemd while using raid0 booting.

Point B.....Readahead seems not viable when using raid striping.

Point C...Of course, using raid0 striping on a single device does not provide any speed nor capacity advantage, but until one has tried it, all performance details are not obvious. Perhaps just being raid0 is enough security for the linux system.

EDIT:11/14/2012.....Suspect that ..run directory, symlinked to /var/run has extended my boot time to ridiculous levels and introduced new elements in the filesystem. The boot time rose from 10 seconds to about 42 seconds on my best CF cards (180MB/s). This occurs in latest kernel as well as previous linux kernels. It may not be due to the run directory but it is certainly a good suspect since it involves booting steps.
Perhaps someone can advise a way to 'cure' this malady?

EDIT: 11/20/2012 Installed raid0 with two maxell 16gb udma devices using 4 partitions on each device for a total of 8 raid devices. It is systemd and boots to xfce4 Desktop in 12 seconds. I assume the latest .iso provides better support for this bootable raid0 config. Grub2
EDIT: 11/21/2012....Utilized hdparm to display the read speed for the new MD11 raid0 combo....................

/dev/sda3:
 Timing buffered disk reads: 260 MB in  3.02 seconds =  86.22 MB/sec

/dev/sda4:
 Timing buffered disk reads: 260 MB in  3.02 seconds =  86.21 MB/sec

/dev/sdb1:
 Timing buffered disk reads: 260 MB in  3.00 seconds =  86.55 MB/sec

/dev/sdb2:
 Timing buffered disk reads: 260 MB in  3.01 seconds =  86.50 MB/sec

/dev/sdb3:
 Timing buffered disk reads: 262 MB in  3.02 seconds =  86.75 MB/sec

/dev/sdb4:
 Timing buffered disk reads: 260 MB in  3.00 seconds =  86.63 MB/sec

/dev/md11:
 Timing buffered disk reads: 530 MB in  3.00 seconds = 176.58 MB/sec
[root@n6re /]#

These are typical speed reads for the maxell 16GB UDMA devices using Sata2CF adapters. The marked rating for the maxell devices is 400x. My mobo supports two primary SATA ports so I cannot increase the devices to three or four to reach the ultimate SataII cap.

The boot time will increase as I add more packages to the singular " root" partition.

This was an experiment to determine the effects of assigning all primary partitions , full capacity(except for 60MB free space for grub2 MBR).

The system could also be formed with single partitioned devices for compaison. These are root only raid0 systems.

EDIT:11:29:2012 Upgraded all raid0 devices to linux-3.6.8-1-x86_64.pkg.tar.xz and all are operating correctly. Have had no failures of any Compact Flash devices.

EDIT: 11:30:2012 Generated systemd xfce4 Desktop in Raid0 pair of CF devices booted in seven seconds. FiveGB of packages. Devices are 266x adata in sata adapters.

EDIT: 12:7:2012....All raid0 devices upgraded to linux-3.6.9-1 and funcitonal.

EDIT: 12/16/12.....Installed 3.6.10-1 via archiso to 16GB maxell UDMA CF card using xfce4 and it boots in 8 seconds to Desktop...3.2GB installed.

EDIT 12/17/12....Generated a four device raid0 on two maxell 16GB UDMA devices using 100 mb for grub-install, 100mb for boot on sda and 100mb for swap on sdb. Two partitions for each device provide the four raid0 items. System boots in seven seconds.
EDIT:12/22/12...Generated a raid0 pair of 16GB devices
EDIT 12/23/12....Installed four 16GB 533x CF cards as bootable raid0 devices in intel 945GCM sata ports(4 ports, 2 master and 2 slave). Boots to xfce4 in seven seconds and hdparm read is 126mB/s. The system size is ~59GB. All four devices need be master for 4 time read speed.
EDIT12/23/12...Installed raid0 bootable to a pair of maxell CF cards 16GB UDMA and obtained 182MB/s and ~20GB system size. All systems are root only raid0 bootable.
EDIT: Jan 12 2013....Install raid0 system to a USB microsd device successfully...boots in 40 seconds.

EDIT; 1/14/2013.....Dscovered special entry for ext4 raid456 in mkinitcpio.conf which enabled ext4 to be installed successfully from arch-install.iso. Entering the ext4 raid456 module in mkinitcpio permits the 3.6.8-1 kernel to properly handle the raid0 params. Result was maxell read speed in raid is 180MB/s with two devices maxell 16GB 400x UDMA compact flash.
EDIT:1/20/13.....Upgraded all raid0 devices to linux3.7.3-1
EDIT: 1/23/13.....Upgraded all raid0 devices to linux3.7.4.1
EDIT: 1/25/13.........DD'ed a video file of 702MB from /var/local1 to/var/local2 at 45MB/s speed with adata16GB pair of Compact Flash devices in raid0. My system is core2 duo Pentium 800mhz Gigabyte 945GCM 2800mhz. Repeating the test with maxell UDMA 400x devices produced a speed of 65MB/s in cp command.

EDIT: 1/26/13....Performed a cp of the archiso of 516MB from Desktop to /var/local2 with results shown :

[root@n6re ~]# time cp /root/Desktop/archlinux /var/local2

real	0m1.189s
user	0m0.013s
sys	0m1.150s
[root@n6re ~]#

This indicates a transfer speed of ~ 216MB/s . This occured with the maxell 400x UDMA pair in raid0. The file transferred is the .iso file for jan 2013 as downloaded from mirror. Next I will try a copy of the dd'ed .iso file as provided on a flash drive after copying it to /rootDesktop.
EDIT: 1/31/13...Edited with world's smallest USB Flash raid0 bootable grub2 device....Kingston micro 8GB....just upgraded to archlinux linux systemd 3.7.5-1.

EDIT:2/1/13....Installed raid0 btrfs in 16GB adata 533X Compact Flash with linux3.7.5-1 as an experiment. Used mdadm_udev and btrfs in Hooks. Set up raid0 in single device with cfdisk and mdadm. Do not know how to create a raid array in btrfs by itself with no mdadm required or even if it is possible. Wiki data is lacking. The fstab for the raid0 array has btrfs as root. Desire also to try compression as an experiment in the raid0 system. No firm info on such a setup has been found.

EDIT:1/2/13: Ran a cp test with maxell 16GB 400X pair and the result:

[root@n6re /]# time cp '/root/Desktop/Patch Adams.avi' /var/local3  (702MB)

real	0m3.145s
user	0m0.020s
sys	0m1.653s
[root@n6re /]#

Speed ~145MB/s

EDIT: 2/8/13....Kingston 600X udma6 raid0 bootable produces 75MB/s in raid dev hdparm and 83MB's for the kingston device alone. Thus, raid0 striping slows the performance but only slightly when the device is used as a single device raid0 array using two partitions. This is surprising for the kingston device which I felt would not boot but works well. Next up is a 600x udma Duracell device at Fry's for $35.
EDIT: 2/10/13:.......Duracell doesn't work. Upgraded all eleven devices in raid0 to linux-3.7.6-1 this date. So far, all devices have performed without failing and none show loss of capacity when operated in root only system with no swap nor home partitions. Some of the devices are five years old. All perform in usb or Sata adapter format as applicable for the device. Smallest is the kingston-micro usb 8GB raid0 bootable grub2. It boots in 45 seconds.

EDIT: 2/13/13.....Upgraded all to linux3.7.7-1.. no problems.
EDIT: 2/16/13...Upgraded to linux-3.7.8-1 on all raid0 devices including the world's smallest raid0 device the kingston micro 8GB.
EDIT: 2/18/13...Upgraded all cf arrays to linux-3.7.9-1 x86_64 with no problems.

EDIT: 2/19/13.. One device is a 16GB adata 533x CF card which is utilizing btrfs in raid0 as an experiment....
EDIT 2/25/13....Upgraded all to linux3.7.9-2-x86_64.pkg.tar.xz

EDIT: 3/3/13....Bought a Monster 500x udma6 CF card at Fry's. It is marked with windows and mac. I installed archlinux raid0 array as per all the rest in that single drive. When attempting to ID the drive with the sata adapter interface it wasn't recognized. I then connected it as USB device and was able to load all packages as desired. It boots however slowly in USB into xfce4 Desktop and performs all functions. One weird thing about the device.
When I install it in the sata adapter while using another device to load archlinux, making it /dev /sdb, it indicates via an led that it is up and ready. However, if I run hdparm -t /dev/sdb, the led is extinguished but the read speed is posted as ...383.30MB/sec !!!!!!..
I am unable to boot from the sata interface with this device and cannot run fsck because it has no superblock and its label is not recognized.

However, it is still a raid0 array and boots from USB into xfce4 Desktop and performs well in archlinux.

Entering blkid in another CF with this device as /sdb gives a uuid for the raid array and cat/proc/mdstat provides the raid0 info for the device as /dev/sdb3 /dev/sdb4 for the single CF card raid array.
I conclude that the device cannot be programmed to accept a superblock.
EDIT 3/6/13....Upgraded all to linux-3.7.10-1-x86_64. Installed aur package cmplayer to all raid0 devices. Nice package!
EDIT: 3/18/13...Upgraded all to systemd 189 and linux3.8.3-2 64 bit with no problems. Awaiting qt5 release which will aid cmplayer to perform even better.
Just one day after St Patties and am still alive with raid0 devices. This jump in linux was unexpected.............
EDIT: 3/23/13...Upgraded to linux-3.8.4-1 on all Raid0 arrays with no problems.
EDIT: 4/4/13.... Upgraded to linux-3.8.5-1 with no problems.
EDIT: 4/15/13...Upgraded twice in archlinux in last few days now am at linux-3.8.7.1 in all raid0 compact flash. No problems encountered. 64bit
EDIT: 4/18/13....Attempting to generate a raid0 system using f2fs filesystem. So far, having difficulty with grub2 bootloader. Not much info to go on for the f2fs-tools and kernel f2fs as to how to get grub2 functioning.
EDIT: 4/22/13.........Upgraded all raid systems to linux3.8.8-1 and cmplayer to .0.7.3-2 No problems in any compact flash devices.
EDIT: 4/28/13....Reinstalled the kingston micro 8GB device with just raid0 grub2 , raid installed to two partitions using the raid root boot for booting and installed linux3.8.10-1 to all raid0 devices. 64bit. All devices operate normally...best boot time to xfce4 desktop is seven seconds with no login nor passwd. The kingston micro is a usb device so it runs at usb speed. With jumanji preloaded urls, the access time is nil to those urls.
EDIT: 5/4/13....Upgraded kingstop usb micro to linux-3.8.11-1-x86_64 and that makes it the simplest raid array I have ...KISS. Only one active partition /root in raid0 bootable. The remaining raid0 compact Flash devices also upgraded.
All devices operate normally, none have failed.
EDIT: 5/12/13.....Upgraded all to linux-3.9.2-1 64bit.
EDIT: 5/20/13....Upgraded all to linux-3.9.3-1 64bit and still unable to establish f2fs uuid /mdxx device. Would enjoy trying the copy-on-write function therein provided in a raid0 boot, root only environment.
EDIT: 5/23/13....Successfully installed f2fs filesystem in raid0 grub2 using latest f2fs-tools from extra repo and using a partitioned grub2 boot. Some of the internet items seem affected in speed of response. Jumanji seems to be normal but firefox is sluggish. A longer time in use will give a better idea of performance. The cf card is 16GB and 533x Adata. I will attempt a different device asap.
EDIT: 5/24/13.....Installed f2fs system in raid0 using Monster 32GB cf card with a partitioned grub2 install. This card does not boot in sata adapter interface but does boot from USB.

Although hdparm is questionable when testing raid arrays, the following results are of interest when running hdparm on the Monster:

[root@n6re /]# hdparm -t /dev/sdb2 /dev/sdb3 /dev/md127

/dev/sdb2:
 Timing buffered disk reads:  32 MB in  3.09 seconds =  10.34 MB/sec

/dev/sdb3:
 Timing buffered disk reads:  32 MB in  3.08 seconds =  10.40 MB/sec

/dev/md127:
 Timing buffered disk reads: 220 MB in  3.03 seconds =  72.71 MB/sec
[root@n6re /]#

This illustrates the ambiguity of array data but it may also indicate an effect of f2fs filesystem differences using copy-on-write procedures. If so, I may have to try it on usb3.0!
The monster cf card is 600x rated.
EDIT 5/24/13....The previous data applies to a sata adapted cf card which makes more sense! My oops.

Connecting the Monster CF card into a 3.0 usb card adapter produces more reasonable answers for hdparm runs:

root@n6re /]# hdparm -t /dev/md126 /dev/sdb2 /dev/sdb3

/dev/md126:
 Timing buffered disk reads:  88 MB in  3.03 seconds =  29.04 MB/sec

/dev/sdb2:
 Timing buffered disk reads:  66 MB in  3.03 seconds =  21.79 MB/sec

/dev/sdb3:
 Timing buffered disk reads:  66 MB in  3.03 seconds =  21.77 MB/sec
[root@n6re /]# hdparm -t /dev/md126 /dev/sdb2 /dev/sdb3

/dev/md126:
 Timing buffered disk reads:  86 MB in  3.01 seconds =  28.58 MB/sec

/dev/sdb2:
 Timing buffered disk reads:  66 MB in  3.04 seconds =  21.71 MB/sec

/dev/sdb3:
 Timing buffered disk reads:  66 MB in  3.03 seconds =  21.76 MB/sec
[root@n6re /]#

It show a modest speed increase with the raid0 array and a better correlation with the component partitions. This speed is good for a usb connected array.

Sorry for the oops! The Monster is not compatible with Sata IDE connect in my machine and is designed for Win.Mac..........
EDIT: 5/25/13... The hdparm result showing 29mB/s is erroneous and is caused by an additional array drive connected in sata adapter while booting into the USB array. Removing the sata drive array changes the hdparm for the usb array to 22 MB/s which is the speed for the partitions of that array. This makes sense out of it'

With a maxell 16gb array having f2fs installed, the hdparm for the array is 68MB/s in the sata adapter which is about 8MB/s less than its two partitions. The array boots in ten seconds to xfce4 Desktop with no log-in nor passwd.. Thus, the raid0 array with a single device has negative effects shown in hdparm but not a serious effect.
Some time will be necessary to evaluate the f2fs effects with raid0...so far so good. That should combine copy-on-write and stripe if all goes well.
I hope to repeat the f2fs raid0 array with two devices rather than a single device.

EDIT: 5/25/13........Generated a bootable f2fs raid0 with two 16MB cf devices...adata 533x cf and Kingston 600x udma.

The hdparm data appears below:

/dev/md127:
Timing buffered disk reads: 260 MB in 3.01 seconds = 86.25 MB/sec (array)

/dev/sda:
Timing buffered disk reads: 198 MB in 3.03 seconds = 65.42 MB/sec (533x)

/dev/sdb:
Timing buffered disk reads: 250 MB in 3.01 seconds = 82.97 MB/sec (600x)

[root@n6re /]#

This data was taken on the running array.

The data indicates the different speeds of the devices.

EdiT:....Performed hdparm on the md127 system while booted into another archlinux raid0 device. Results below indicate raid performance present:

[root@n6re ~]# hdparm -t /dev/md127 /dev/sda /dev/sdb

/dev/md127:
 Timing buffered disk reads: 386 MB in  3.00 seconds = 128.47 MB/sec

/dev/sda:
 Timing buffered disk reads: 196 MB in  3.03 seconds =  64.71 MB/sec

/dev/sdb:
 Timing buffered disk reads: 222 MB in  3.01 seconds =  73.83 MB/sec
[root@n6re ~]#

I cannot increase the number of devices at this time. This data indicates raid0 is operative but isn't necessarily accurate as is often stated for raid devices. However, the speeds for the two devices are in line with their capabilities and I suspect the raid number is reasonable. This system in md127 has f2fs filesystem installed.[root@n6re /]# systemd-analyze
Startup finished in 1.831s (kernel) + 1.451s (userspace) = 3.283s
[root@n6re /]#

Systemd-analyze results for /dev/md127 boot:
[root@n6re /]# systemd-analyze
Startup finished in 1.831s (kernel) + 1.451s (userspace) = 3.283s
[root@n6re /]#
EDIT: 5/28/13...Upgraded to linux-3.9.4-1 in two f2fs CF device raid0 and in single device f2fs CF raid0. All performing normally.

EDIT5/28/13 ....Installed f2fs to maxell CF cards (2) and the latest linux-3.9.4-1 was installed. The boot to xfce4 Desktop is seven seconds. Hdparm from another device boot is as follows:

[root@n6re ~]# hdparm -t /dev/md127 /dev/sda2 /dev/sdc2

/dev/md127:
 Timing buffered disk reads: 540 MB in  3.00 seconds = 179.93 MB/sec

/dev/sda2:
 Timing buffered disk reads: 244 MB in  3.01 seconds =  81.17 MB/sec

/dev/sdc2:
 Timing buffered disk reads: 246 MB in  3.02 seconds =  81.57 MB/sec
[root@n6re ~]# hdparm -t /dev/md127 /dev/sda2 /dev/sdc2

/dev/md127:
 Timing buffered disk reads: 540 MB in  3.00 seconds = 179.88 MB/sec

/dev/sda2:
 Timing buffered disk reads: 244 MB in  3.00 seconds =  81.25 MB/sec

/dev/sdc2:
 Timing buffered disk reads: 246 MB in  3.01 seconds =  81.65 MB/sec
[root@n6re ~]#  

;code]

Thus the maxell 400x udma card pair wins the speed prize. I have no room for any more Sata ports nor can I find more maxell CF devices on the market, altho I haven't a place to employ them anyhow.

EDIT: 6/9/13....Timed cp of 60mb video file from /tmp to /Desktop occured at 432mb/s after initially placing the file into /tmp from /root. I assume it used cache during the transfer. Original cp from /root into /tmp occured at 76mb/s. This activity performed in f2fs filesystem as installed in raid0 in two 8GB 266x slc adata cf cards.
All raid systems upgraded to new /usr/bin accomodation.
As expected, the performance of slc devices is generally better than mlc cards.
EDIT: 6/15/13....Installed archlinux .iso with f2fs in raid0 using maxell 16GB CF cards using partitioned grub2 as per grub wiki chattr procedure. Utilized /dev/sda1 as boot partition and partitioned /dev/sdb entirely as /dev/sdb1. The resulting install was bootable and the post=install to systemd successful. I suspect that this attempt resulted in success because it utilized the new linux-3.9.6-1 kernel release. Previous attempts to install a bootable pair of CF cards failed with previous linux kernel releases. So the system is raid0 f2fs filesystem bootable from grub2 partition.
Strange indeed that release 6-1 occured on my wedding anniversary this date...anniversary number 66.
EDIT: 6/23/13...Upgraded all devices to linux-3.9.7-1 including f2fs systems, a pair of two CF device f2fs raid0 and a single device Cf f2fs ad one sdhc single device with f2fs raid0. All systems running normally.

EDIT: 7/4/13....Generated raid0 bootable f2fs system with 2 silicon power 32GB compact Devices (600x udma5) in linux-3.9.8-1 kernel . Posting this in that system.

EDIT: 7/6/13....Upgrade to linux3.9.9-1 broke the raid0 pairs in f2fs mode. Sdhc single device raid0 still functions correctly in the upgrade.

Decided to wait for f2fs upgrade in grub before pursuing any further raid arrays in f2fs with compact flash devices. Sdhc may well be the only useable pairs at this time. I may try a pair of 8gb devices but they are usb only, perhaps 40 mb/s max in raid0.
EDIT: 7/10/13.....Ran systemd-analyze on a 32GB raid0 pair of CF devices with the results shown" 1.484 kernel and userspace 1.301....xfce4 Desktop in seven seconds. Power off occurs in two seconds.

EDIT: 7/21/13......Re-installed f2fs filesystem to raid0 pair of maxell 16GB 400x udma Compact Flash devices.
Booting this raid0 array from a third device connected via USB which has the same kernel and f2fs bootable via partitioned f2fs grub2 using the wiki procedure for grub partitioned boot. The raid0 array has no bootloader data entered and operates from Master Sata port using a compact Flash adapter. Raid array boots from the USB f2fs device by default. Linux-3.9.9-1 kernel in raid array and the booting device in USB. The array boots into xfce4 Desktop in seven seconds and is root only. I post this from that setup.

EDIT: 7/24/13..Upgraded all successfully to linux-3.10.2-1 including the worlds smallest raid0 device. Several f2fs devices also upgraded but still need assist in booting. So far, no failures in any CF devices.
EDIT: 8/23/13...Upgrade success to linux-3.10.7-1....
EDIT: 8/28/13....Upgrade to latest linux successful with bootable f2fs CF pair :
[root@n6re ~]# uname -a
Linux n6re 3.10.9-1-ARCH #1 SMP PREEMPT Wed Aug 21 13:49:35 CEST 2013 x86_64 GNU/Linux
[root@n6re ~]#

This pair of CF cards boots f2fs without an auxiliary helper. I continue evaluating that fs in compact flash.
EDIT: 8/31/13..............Upgraded all devices to linux-3.10.10-1-x86_64.pkg.tar.xz. No problems...
EDIT: 9/8/13...F2FS devices now fail in latest linux upgrade. Reason is unknown at this time. Ext4 versions all ok. Suspect xf86 upgrades.
EDIT: ....9/11/13...Attempting to follow the grub data for partitioned boot and find it fails as presently provided in the grub wiki. It seems it is out of sequence and also missing critical steps to enable partitioned boot. When properly utilized it ends up with incorrect uuid which seems to be generated by one of four packages, grub itself,udisks2, systemd, or mdadm. The uuid generated is different from the genfstab provided in the install sequence and the boot defaults to grub rescue>. Examination of the raid array in cat /proc/mdstat says it is active but it cannot boot with incorect uuid. I googled and found one statement that says grub automatically uses uuid which might explain the different uuid. All f2fs attempts to boot now fail in raid0 systems. All ext4 systems operate normally.
EDIT: 9/16/13.....Re-installed two pairs of maxell 16GB 400x CF cards with f2fs filesystems raid0. Hdparm reports 176MB/s and 181MB/s for the raid arrays(2). Each array requires a helper f2fs filesystem to boot. Haven't found a solution to that problem yet.
EDIT:...9/17/13....Upgraded several systems to linux-3.11.1-1 but some are in limbo with grub problems. May have to change from blocklists soon due to the rate of change in linux.

EDIT:..9/21/13.....Installed Btrfs into a pair of maxell 400x udma CF cards using the command:

 btrfs -m raid0 /dev/sda /dev/sdb

.
Desired to verify the existence of btrfsraid and searched the net for a means of doing such a feat. Found a
method......

 btrfs fi df /mountpoint

I mounted the btrfs system in /mnt/md with:

 mount /dev/sda /mnt/md

I then entered: btrfs fi df /mnt/md..............

This command produced the following verification data for raid0:

 [root@n6re Desktop]# btrfs fi df /mnt/md
Data, RAID0: total=6.00GB, used=4.26GB
Data: total=8.00MB, used=4.51MB
System, RAID0: total=16.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, RAID0: total=2.00GB, used=135.93MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]#

This raid0 is not md raid but provides striped activity on the raid0 pair established by the btrfs install. AFAICT this is the only way to verify directly a correct install to btrfs raid0.
I do not have capability in my mobo to use more than two CF cards with Sata interface.
Hdparm -t reports 85 MB/s for both /sda and /sdb. I have yet to find a way to read the raid speed. I did run a timed cp of a 1gb video file and obtained a speed of 190+MB/s. This is consistent with the expected level near 170 - 180 MB/s.

However the boot time increased to 20 seconds with btrfs raid0. There are some error displays during boot but they do not stop the full boot process but do delay it.
The system installs to full-capacity Compact Flash devices with no partitioning required. The devices were wiped with dd before the install via archlinux.iso. The normal grub install procedure below was used:

 grub-install --recheck /dev/sda 
grub-mkconfig -o /boot/grub/grub.cfg
mkinitcpio -p linux

So far...so good. I acknowledge e-mail help and succor from wonderwoofy on the archlinux forums, thanks again Curtis.

Again, btrfs raid is not md raid and is handled differently from that mode.
In my setup, it is root only.

EDIT: 9/24/13..Tried a four device raid0 but was un[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]# [root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#
[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127
[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

[root@n6re Desktop]#

[root@n6re Desktop]# btrfs fi df /[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

[root@n6re Desktop]#

Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda
[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

System, DUP: total=8.00MB, used=4.00KB[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

dev/md127
[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

successfull. I re-installed a raid0 to three devices and obtained the following:

[root@n6re Desktop]# btrfs fi df /
Data, RAID0: total=6.00GB, used=4.03GB
Data: total=8.00MB, used=0.00
System, RAID0: total=15.94MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, RAID0: total=3.00GB, used=144.39MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]#

If I use the balance procedure, the third device is removed from the raid but remains as an added drive. So I plan to leave the arrangement as is and work with it.
It may be interesting when the upgrade to next linux occurs!!
EDIT: 9/25.13.......Upgrade to linux-3.11.1-2 successfully installed and running. Not a major upgrade, included udisks2 ..so far so good.

May the bluebird of happiness twiddle your bits!!!

EDIT 9/25/13.Re-installed four devices with btrfs after elimuinating old data from fourth devcie. This was successful and all devices report 85MB/s read speed in hdparm. Btrfs pertinent data follows:

[root@n6re Desktop]# btrfs fi df /
Data, RAID0: total=8.00GB, used=4.06GB
Data: total=8.00MB, used=0.00
System, RAID0: total=16.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, RAID0: total=4.00GB, used=146.75MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]

[root@n6re Desktop]# btrfs filesystem show
Label: none  uuid: d88851a1-e7f2-4e10-971b-02feaffe785a
	Total devices 4 FS bytes used 4.20GB
	devid    4 size 14.89GB used 3.00GB path /dev/sdd
	devid    3 size 14.89GB used 3.00GB path /dev/sdc
	devid    1 size 14.89GB used 3.02GB path /dev/sda
	devid    2 size 14.89GB used 3.00GB path /dev/sdb

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

Thus I am able to utilize all four ports of my SataII with CF adapters if I run btrfs on full capacity devices.

The boot time is ~20 seconds and capacity is 64GB raid0 btrfs.

EDIT: .../9/27/13.....Installed btrfs to a three device MDRAID array using free space 100mb on sda and a full sized sda1 partition along with full sized /dev/sdb and /dev/sdc. Total size for the three adata 16gb 533x CF cards is 48MB. The install was with arch.iso and grub was installed with:.... grub-install /dev/sda...followed by grub-mkconfig and mkinitcpio -p linux.

The raid array is named md127 and it boots from sda in ~9 seconds.

Pertienent data follows:[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.71GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=133.31MB
Metadata: total=8.00MB, used=0.00

The system filesystem dats:
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: d952a434-ee3d-40cd-a654-f66fcfc92e71
Total devices 1 FS bytes used 1.84GB
devid 1 size 44.67GB used 4.04GB path /dev/md127

Label: none uuid: 2fdaabf7-449c-48ad-afe1-78fc683e6be8
Total devices 1 FS bytes used 28.00KB
devid 1 size 14.92GB used 2.04GB path /dev/sda

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#
As is shown, the system is md127 raid array and responds to btrfs queries. The boot time is quick to Desktop.

The performance in hdparm -t shows some increase in read speed in the raid array listing but this is being run in a mounted array:
/dev/md127:
Timing buffered disk reads: 286 MB in 3.03 seconds = 94.50 MB/sec

/dev/sda:
Timing buffered disk reads: 200 MB in 3.02 seconds = 66.15 MB/sec

/dev/sdb:
Timing buffered disk reads: 188 MB in 3.02 seconds = 62.24 MB/sec

/dev/sdc:
Timing buffered disk reads: 194 MB in 3.01 seconds = 64.50 MB/sec
[root@n6re Desktop]#

Just did this so performance will be studies for some time.

I plan to implement this in a four device array soon.

EDIT: ...9/27.13.....Successfully installed mdraid to four btrfs devices using similar technique with grub.

The pertinent data follows:

[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.87GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=134.14MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]#

And the following:
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: ec3ec36e-ad9a-43af-b739-3720bbba1a9f
Total devices 1 FS bytes used 2.00GB
devid 1 size 59.46GB used 4.04GB path /dev/md127

Btrfs v0.20-rc1-253-g7854c8b
[root@n6re Desktop]#

I do not have boot time as yet for this config...........EDIT: ~8seconds to xfce4 Desktop

Strange details for this one!!! EDIT: Hdparm data indicates only one pair exibits raid performance, as shown:
root@n6re Desktop]# hdparm -t /dev/md127 /dev/sda /dev/sdb /dev/sdc /dev/sdd

/dev/md127:
Timing buffered disk reads: 544 MB in 3.00 seconds = 181.10 MB/sec

/dev/sda:
Timing buffered disk reads: 258 MB in 3.00 seconds = 85.91 MB/sec

/dev/sdb:
Timing buffered disk reads: 260 MB in 3.01 seconds = 86.28 MB/sec

/dev/sdc:
Timing buffered disk reads: 260 MB in 3.00 seconds = 86.61 MB/sec

/dev/sdd:
Timing buffered disk reads: 258 MB in 3.02 seconds = 85.47 MB/sec
[root@n6re Desktop]#

These are maxell 400x udma 16GB devices and I cannot find them anywhere anymore!!!!

EDIT: 928/13..........Removed the four device system and re-installed the paired device btrfs because the use of four devices is not compatible with my mobo. A four device setup requires four independent Sata ports. I am not sure such a mobo is manufactured..most all have only three ndependent ports. A limitation of the controllers I assume.

I now have a two device system in btrfs and a two device system in ext4. I have eliminated the mdadm based system of four devices which did not provide four times read speed but only two times.

The four device seemed possible with four ports and four separate cables but the controller is king!!!

EDIT: 10/4/13..Created btrfs pair with 8GB a data 266x cf devices. Ran dd on both devices to clear. Installed to full device capacity with the following command in arch .iso procedure: (no partitions)....

mkfs.btrfs -m raid0 /dev/sda /dev/sdb............

Followed this with the command:

mount /dev/sda /mnt.........

Then pacstrap step follows.

In the arch-chroot after setting up the params, pacman installed grub.

Grub was installed with:

grub-install /dev/sda...

This was successful.

Grub-mkconfig -o /boot/grub/grub.cfg followed.

This produced a grub error statement which was ignored.

Enter mkinitcpio -p linux.

Rebooted into /dev/sda and received the login prompt and proceeded to install pacman packages successfully.

This produces btrfs filesystem and operates two full-sized compact flash devices un-partitioned and operates in root mode.

I post this edit in that btrfs pair boot-up.

This is the simplest KISS I have encountered in Linux. No need for cfdisk, gparted, fdisk or any partitioning.
EDIT: 10/6/13.Successfully upgraded to btrfs linux-3.11.3-1 on five pairs of cf raid0 cards. Best boot time is six seconds to xfce4 Desktop. Most btrfs pairs are full capacity installs with grub bootloader.

EDIT: 10/6/13...No sooner got linux-3.11.3-1 installed and found linux-3.11.4-1 to install. Successfully installed btrfs in all devices full capacity with grub bootloader.

EDIT: 10/7/13...Some data of interest with the btrfs device raid0:
Grub.cfg detail:
### BEGIN /etc/grub.d/10_linux ###
menuentry 'Arch Linux, with Linux core repo kernel' --class arch --class gnu-linux --class gnu --class os $menuentry_id_option 'gnulinux-core repo kernel-true-/dev/sda
/dev/sdc' {
load_video
set gfxpayload=keep
insmod gzio
insmod btrfs
set root='hd0'
if [ x$feature_platform_search_hint = xy ]; then
search --no-floppy --fs-uuid --set=root --hint-bios=hd0 --hint-efi=hd0 --hint-baremetal=ahci0
--hint-bios=hd2 --hint-efi=hd2 --hint-baremetal=ahci2 c14a83bf-6ee7-4311-bba1-ad9ee8e960a2
else
search --no-floppy --fs-uuid --set=root c14a83bf-6ee7-4311-bba1-ad9ee8e960a2
fi
echo 'Loading Linux core repo kernel ...'
linux /boot/vmlinuz-linux root=UUID=c14a83bf-6ee7-4311-bba1-ad9ee8e960a2 rw quiet
echo 'Loading initial ramdisk ...'
initrd /boot/initramfs-linux.img
}
The grub data shows the single device uuid in all instances signifying no partitions are present and the unusual set root statement indicates full sized device use.

Further data shows the uuid applies to two full sized drives:

[root@localhost Desktop]# blkid
/dev/sda: UUID="c14a83bf-6ee7-4311-bba1-ad9ee8e960a2" UUID_SUB="3ae14a08-cdea-4f97-a038-50178373c425" TYPE="btrfs" PTTYPE="dos"
/dev/sdd: UUID="16504ad5-dc88-588f-6a79-30d116bd0c4b" UUID_SUB="e2245352-3ec5-a53c-47f8-93c8b35f4748" LABEL="archiso:40" TYPE="linux_raid_member"
/dev/sdb1: UUID="16504ad5-dc88-588f-6a79-30d116bd0c4b" UUID_SUB="4c5ed737-498a-b5fb-1e77-384f6e010e9e" LABEL="archiso:40" TYPE="linux_raid_member"
/dev/md127: UUID="1d32a445-5003-4f62-9e60-58600479e7d2" TYPE="ext4"
/dev/sdc: UUID="c14a83bf-6ee7-4311-bba1-ad9ee8e960a2" UUID_SUB="44880ced-9de7-4a68-a007-965596bf1de0" TYPE="btrfs"
[root@localhost Desktop]#

The display above shows /dev/sda and /dev/ sdc as one pair and /dev/sdb/and /dev/sdd as a second pair with each pair having the same uuid> These devices are mounted one pair in Master and the other in Slave to permit two instances of btrfs to be accessible for booting in bios select mode.

The running system has the following:
[root@localhost Desktop]# btrfs fi df /
Data, RAID0: total=2.00GB, used=1.94GB
Data: total=8.00MB, used=0.00
System, RAID0: total=16.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, RAID0: total=2.00GB, used=134.54MB
Metadata: total=8.00MB, used=0.00
[root@localhost Desktop]#

Additional data:

[root@localhost Desktop]# btrfs filesystem show
Label: none uuid: c14a83bf-6ee7-4311-bba1-ad9ee8e960a2
Total devices 2 FS bytes used 2.07GB
devid 2 size 14.92GB used 2.01GB path /dev/sdc
devid 1 size 14.92GB used 2.03GB path /dev/sda

Btrfs v0.20-rc1-358-g194aa4a
[root@localhost Desktop]#

These data are for the latest linux-3.11.4-1.

Be aware that btrfs raid0 is not mdadm raid0

EDIT: 10/10/13.....Installed btrfs to full-size kingston usb micro device 8GB. Functions nicely but usb speeds. This is world's smallest btrfs single device system.

EDIT:10/11/13....With 32GB Silicon Power cf cards the start up times are:

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.305s (kernel) + 2.246s (userspace) = 3.552s
[root@n6re Desktop]#
This is in btrfs filesystem and booting directly to xfce4 background pix...no icons...right click to open menu....background ix is hot-air balloon soaring into blue sky over Park City, Utah...up,up, and away!!!! Good theme!!

EDIT: 10/13/13 ....Ran analyze on pair of 32GB CF devices in btrfs each device is 600x as marked Silicon Power but only show udma5....

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.163s (kernel) + 1.953s (userspace) = 3.117s
[root@n6re Desktop]#

Very fine result for MLC device.

I have added compression=izo to fstab.

The system reaches xfce4 Desktop in ~13 seconds. I may find a way to reduce that in btrfs since an ext4 reaches Desktop in six seconds.
EDIT: 10/14/13 ....Upgraded all raid0 systems to linux-3.11.5-1 with no problems.

EDIT: 10/18/13...Reinstalled btrfs to a pair of 32GB 600x Silicon Power devices . I used mkfs.btrfs -d raid0 /dev/sda /dev/sdb. The pertinent data on this install follows:

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.199s (kernel) + 1.575s (userspace) = 2.774s
[root@n6re Desktop]#

Also:

[root@n6re Desktop]# btrfs fi df /
Data, RAID0: total=4.00GB, used=2.03GB
Data: total=8.00MB, used=8.00MB
System, RAID1: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, RAID1: total=1.00GB, used=140.81MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]#

And, lastly:

[root@n6re Desktop]# btrfs filesystem show
Label: none  uuid: 87ec57fe-48b2-4470-a5e9-aceaad6b76dd
	Total devices 2 FS bytes used 2.18GB
	devid    2 size 29.82GB used 3.01GB path /dev/sdb
	devid    1 size 29.82GB used 3.03GB path /dev/sda

Btrfs v0.20-rc1-358-g194aa4a
[root@n6re Desktop]#

The install includes thunderbird, firefox,cmplayer,vlc,jumanji,xfce4.alsa-utils,flashplugin gparted,hdparm and feh. System is x86_64 in gcm945 S2 mobo using sata to CF card adapters. Boot is direct to xfce4 desktop with no login required.

EDIT: More info on the 32GB pair:

  [root@n6re Desktop]# systemd-analyze
Startup finished in 1.137s (kernel) + 2.442s (userspace) = 3.579s
[root@n6re Desktop]#

EDIT: 10/20/13......Again, the boot time for the 32GB devices in btrfs raid after upgrade to 3.11.6-1:

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.098s (kernel) + 2.575s (userspace) = 3.673s
[root@n6re Desktop]#

EDIT: 10/23/13..........Upgraded maxell 400x 16GB udma pair of CF btrfs mode devices with the following results:

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.194s (kernel) + 1.490s (userspace) = 2.684s
[root@n6re Desktop]#

Linux-3.11.6-1 x86_64

EDIT 1029/13....Examined ZFS to determine what I might do with that system. Found to my dismay it doesn't upgrade without reinstall. Wanted to give raid0 a try in that system but gave it up due to upgrade mess. Btrfs also doesn't need a partition for booting, at least, not one with that name and procedures to prepare....like ZFS does. Btrfs performs very well in archlinux with boot times as low as 5 seconds in systemd-analyze and 2 seconds poweroff. So for now no ZFS.

EDIT: Halloween 10/31/13....Install btrfs to single drive adata 8GB. Used command ...mkfs.btrfs -d single /dev/sdb.....The btrfs data and systemd-analyze follow for the small system with xfce4 Desktop as performed in all units discussed:

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.257s (kernel) + 4.124s (userspace) = 5.381s
[root@n6re Desktop]# btrfs filesystem show
Label: none uuid: 16d8f2c9-84c2-48ac-b7b9-21273d0eb0e4
Total devices 1 FS bytes used 1.58GB
devid 1 size 7.59GB used 1.80GB path /dev/sdb

Btrfs v0.20-rc1-358-g194aa4a
[root@n6re Desktop]# btrfs fi df /
Data: total=1.54GB, used=1.46GB
System: total=4.00MB, used=4.00KB
Metadata: total=264.00MB, used=118.03MB
[root@n6re Desktop]#

Just an experiment with no known advantages in the method.

EDIT: 11/9/13..Performed the procedure outlined in btrfs wiki to eliminate the error during boot as it appears in all the systems I have installed. The procedure removed the error during boot for all the btrfs systems(remove (") from both ends of echo statement).

Also performed successful use of defrag on one btrfs system and plan to do it on all remaining pairs. This is accomplished with the command.....btrfs fi defragment /........ and it defrags /root which in my case is both full-capacity Compact Flash devices in the btrfs pair.

Searching for additional CF devices with UDMA and not too expensive since I need at least two to do raid0. The market seems to have dried up with the phone pix more popular.

EDIT: ..11/9/13... Completed the defrag on all pair successfully.

Performed an add procedure on an 8GB adata pair in btrfs.

First, installed a third CF device in an IDE slot of the motherboard, whreas the other two devices are in Sata/CF adapters. This device is ID'ed as /sev/sda. Ran dd on it to clear it.

Then performed the following:

mount /dev /sdb /mnt (mounts the btrfs pair in /mnt)
btrfs device add /dev/sda /mnt

Follow this with:

btrfs filesystem balance start /mnt

The procedure produced the following results:

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.259s (kernel) + 3.207s (userspace) = 4.467s
[root@n6re Desktop]#

[root@n6re Desktop]# btrfs fi df /
Data, RAID0: total=4.59GB, used=2.43GB
System: total=4.00MB, used=4.00KB
Metadata, RAID0: total=288.00MB, used=140.91MB
[root@n6re Desktop]#

The boot time shown is less than half of the normal boot time for the original pair /dev/sdb /dev/sdc.

The time displayed is the time for the btrfs system trio to reach xfce4 Desktop with background pix from the grub prompt.

The entire system operates in root mode only and i enter this edit with that operating system.

EDIT: 11/10/13......A note describing the disparity between IDE performance and Sata adapter IDE performance using hdparm (/dev/sda is IDE port and the others are Sata/IDE ports)

[root@n6re Desktop]# hdparm -t /dev/sda /dev/sdb /dev/sdc

/dev/sda:
Timing buffered disk reads: 202 MB in 3.02 seconds = 66.86 MB/sec

/dev/sdb:
Timing buffered disk reads: 260 MB in 3.02 seconds = 86.19 MB/sec

/dev/sdc:
Timing buffered disk reads: 256 MB in 3.00 seconds = 85.31 MB/sec
[root@n6re Desktop]#

All three are maxell 400x UDMA devices. This test was performed on a three device btrfs raid0 system booting in ~ 7 secs. System operates in root mode only. Total capacity is 48 MB with full device capacity on the three drives

EDIT: 11/10/13......Generated a three device btrfs raid0 system with two 16GB CF devices and one 8GB device. The system boots in 7.5 secs and powers off in 2 secs. The arrangement allows for differing capacities of the elements of the btrfs raid0 system. I now have four three-device btrfs raid0 systems, each boots in 7.5 secs and powers off in 2 sec. All are root only full capacity device arrangements.

EDIT: 11/10/13...Ran hdparm on the dissimilar sized three device system with /dev/sda in the IDE slot and the others in Sata/adpters...

[root@n6re Desktop]# hdparm -t /dev/sda /dev/sdb /dev/sdc

/dev/sda:
Timing buffered disk reads: 118 MB in 3.04 seconds = 38.87 MB/sec

/dev/sdb:
Timing buffered disk reads: 188 MB in 3.01 seconds = 62.47 MB/sec

/dev/sdc:
Timing buffered disk reads: 196 MB in 3.03 seconds = 64.69 MB/sec
[root@n6re Desktop]#

This result provides a different arrangement of the /dev/sda because when utilizing it in a sata slot, the system cannot boot. It always boots from the IDE slot in the mobo in the btrfs triple device arrangement.

EDIT: 11/11/13....Kingston 16GB 600x in /dev/sda and two Adata 16gb 533x /dev/sdb /dev/sdc with boot on /dev/sdb produced the following upon booting this morning:

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.111s (kernel) + 3.454s (userspace) = 4.566s
[root@n6re Desktop]#

I find that the times are not necessarily constant. This result on three devices using the full capacity of all devices........

[root@n6re Desktop]# btrfs fi show
Label: none uuid: 61f6e2f2-18ee-4c70-b322-75d60554de22
Total devices 3 FS bytes used 2.69GB
devid 2 size 14.92GB used 1.28GB path /dev/sdb
devid 1 size 14.92GB used 1.04GB path /dev/sdc
devid 3 size 14.89GB used 1.25GB path /dev/sda

Btrfs v0.20-rc1-358-g194aa4a
[root@n6re Desktop]#

EDIT: 11/13/13....Installed a pcie interface to CF card with addonics adapter to enable a four device btrfs arrangement. Utilized ...mkfs.btrfs -d raid10 with four devices named. The result was not very good. The speeds of the devices were badly affected and the systemd-analyze was very poor. It is shown below:

[root@n6re Desktop]# systemd-analyze
Startup finished in 6.335s (kernel) + 1min 30.100s (userspace) = 1min 36.435s
[root@n6re Desktop]#

I will attempt this again but will be more attentive to the btrfs command for raid10. It is also possible that the use of pcie impacts the result negatively. I post from the four device setup.

EDIT: 11/14/13....Re-installed the three device maxell 400x udma devices.

Th result of this change is....... WHAT? Data show below:

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.178s (kernel) + 638ms (userspace) = 1.817s
[root@n6re Desktop]#

Wha hoppen?

EDIT: Repeated the systemd-analyze process with identical result!!!! After reboot.............

EDIT: After installing cmplayer, received the following result from systemd..

root@n6re Desktop]# systemd-analyze
Startup finished in 1.180s (kernel) + 5.500s (userspace) = 6.680s
[root@n6re Desktop]

[root@n6re Desktop]# btrfs fi df /
Data, RAID0: total=3.00GB, used=2.07GB
Data: total=8.00MB, used=0.00
System, RAID1: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, RAID1: total=1.00GB, used=145.46MB
Metadata: total=8.00MB, used=0.00

[root@n6re Desktop]# btrfs fi show
Label: none uuid: f05a9ddf-86f0-497c-96c9-7150888082f1
Total devices 3 FS bytes used 2.21GB
devid 2 size 14.89GB used 1.01GB path /dev/sdb
devid 1 size 14.89GB used 2.02GB path /dev/sda
devid 3 size 14.89GB used 2.01GB path /dev/sdc

Btrfs v0.20-rc1-358-g194aa4a-dirty
[root@n6re Desktop]#

[root@n6re Desktop]#

EDIT: .....Upgrade to 3.12-1 linux successful with the following 3 device analyze:

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.764s (kernel) + 4.446s (userspace) = 6.210s
[root@n6re Desktop]#

Have four more triple systems to upgrade....

EDIT: ...Last 16GB triple produced a 4.653sec result. A 32 GB triple produced 15.145 sec result. An 8GB triple produced a 7.777sec result and a maxell 16GB triple produced 13.766 sec result.
I assume these numbers are the result of differing order of units in systemd in addition to the actual size of the triple.

Of interest, the low figure for the last 16GB triple made up of two 16GB devices and an 8GB device(therefore not a triple 16) the details are shown below:

[root@n6re Desktop]# btrfs fi df /
Data, RAID0: total=18.00GB, used=14.15GB
Data: total=8.00MB, used=1.56MB
System, RAID1: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, RAID1: total=1.00GB, used=195.25MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs fi show
Label: none uuid: d82d69ea-accb-4906-adb0-a724e740c59e
Total devices 3 FS bytes used 14.34GB
devid 2 size 14.92GB used 6.01GB path /dev/sdb
devid 1 size 7.51GB used 7.02GB path /dev/sda
devid 3 size 14.92GB used 7.01GB path /dev/sdc

Btrfs v0.20-rc1-358-g194aa4a-dirty
[root@n6re Desktop]#
It is seen as raid1. The ID says "dirty" which it may be!!!!

EDIT: 11/14/13/...AFAICT the use of systemd-analyze does not play well with btrfs. The results of sequentially rebooting and testing systemd give wild answers...8 secs blame for log-ind.service and then none!!! Perhaps this is caused by btrfs raid0 which is not the same as md-mdadm raid.

EDIT: 11/16/13...Performed a delete function on a three device maxell raid0 btrfs array. I post the method below while on the deleted system:

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.111s (kernel) + 12.591s (userspace) = 13.702s
[root@n6re Desktop]# mount /dev/sdb /mnt
[root@n6re Desktop]# btrfs device delete /dev/sda /mnt

[root@n6re Desktop]#
[root@n6re Desktop]#

The systemd time should reduce greatly after this activity.

EDIT: 11/17/13....The experiment failed to boot...probably because i deleted wrong device!

Today, I tried a new tack....make a single device btrfs and examine its performance.

Iutilized a maxell 400x udma 16GB device and set-up in archiso install as follows:

mkfs.btrfs -f -m single /dev/sda....( -f to remove old btrfs data)

This permitted the install of archlinux and post-install of most of my desired packages.

I then ran the system from fallback grub and obtained the following results:

root@n6re Desktop]# systemd-analyze
Startup finished in 1.644s (kernel) + 2.449s (userspace) = 4.093s
[root@n6re Desktop]# btrfs fi df /
Data: total=3.01GB, used=2.49GB
System: total=4.00MB, used=4.00KB
Metadata: total=520.00MB, used=199.61MB
[root@n6re Desktop]# btrfs fi show
Label: none uuid: 9602b847-92b1-448e-a142-91a148ea0ba0
Total devices 1 FS bytes used 2.69GB
devid 1 size 14.89GB used 3.52GB path /dev/sda

Btrfs v0.20-rc1-358-g194aa4a-dirty
[root@n6re Desktop]#

The figures above were obtained for boot into xfce4 gui Desktop with background jpeg picture

The performance in normal grub boot was 10 secs in analyze. The blame in arch boot was 8 secs for alsa-restore.service.

The figures for fallback grub were obtained after removing alsa-utils.

I suspect I may be able to generate an md raid using the single btrfs mode to generate a fs on md raid as a single target.

So a single device can perform very snappily in btrfs single mode. Raid0 in btrfs may not be any better. I haven't any data for partitioned btrfs raid0 because I suspect I cannot boot it with grub

I install grub in this single in a normal grub install sequence with no errors. The device used is as a full-capacity device of 14.89GB. So grub-mkconfig is applied via btrfs-progs without a partition.

EDIT: Repeated a single device btrfs install with the command ....mkfs.btrfs -d single /dev/sda...and successfully installed and post-installed my packages. The first trial of the performance data on grub arch produced the following:

Startup finished in 1.211s (kernel) + 5.217s (userspace) = 6.428s
[root@n6re Desktop]#
[root@n6re Desktop]#
[root@n6re Desktop]#
[root@n6re Desktop]# btrfs fi df /
Data: total=2.01GB, used=1.79GB
System, DUP: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, DUP: total=1.00GB, used=135.16MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]#
[root@n6re Desktop]#
[root@n6re Desktop]#
[root@n6re Desktop]# btrfs fi show
Label: none uuid: 65980579-6028-4845-adf5-9b91d56a7957
Total devices 1 FS bytes used 1.92GB
devid 1 size 14.89GB used 4.04GB path /dev/sdb

Label: none uuid: 9602b847-92b1-448e-a142-91a148ea0ba0
Total devices 1 FS bytes used 2.84GB
devid 1 size 14.89GB used 3.52GB path /dev/sda

Btrfs v0.20-rc1-358-g194aa4a-dirty
[root@n6re Desktop]#

As is shown, /dev/sda and /dev/sdb are both single devices of similar size, sda is -m single and sdb is -d single.

I will be testing these two for some time.

EDIT: 11/18/13...Decided to try a wild idea...With both single devices in place as sata adapted cf cards...booted on device /sda (m derived single device btrfs) and mounted /dev/sda in /mnt/md.

Then ran....btrfs device add /dev/sdb /mnt/md.....

I detected no change in /dev/sdb and the procedure exited after a few seconds.
Subsequently, could not boot on /dev/sdb. Also could not boot on /dev/sda.

Therefore, I needed both devices to complete booting indicating that the two devices were now a btrfs pair. The results are as follows:

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.162s (kernel) + 12.216s (userspace) = 13.379s
[root@n6re Desktop]# btrfs fi show
Label: none uuid: 9602b847-92b1-448e-a142-91a148ea0ba0
Total devices 2 FS bytes used 2.87GB
devid 2 size 14.89GB used 0.00 path /dev/sdb
devid 1 size 14.89GB used 3.52GB path /dev/sda

Btrfs v0.20-rc1-358-g194aa4a-dirty
[root@n6re Desktop]# btrfs fi df /
Data: total=3.01GB, used=2.67GB
System: total=4.00MB, used=4.00KB
Metadata: total=520.00MB, used=203.55MB
[root@n6re Desktop]# df -h
Filesystem Size Used Avail Use% Mounted on
/dev/sda 30G 2.9G 27G 10% /
dev 1.6G 0 1.6G 0% /dev
run 1.6G 328K 1.6G 1% /run
tmpfs 1.6G 0 1.6G 0% /dev/shm
tmpfs 1.6G 0 1.6G 0% /sys/fs/cgroup
tmpfs 1.6G 8.0K 1.6G 1% /tmp
[root@n6re Desktop]# mount /dev/sdb /mnt/md
[root@n6re Desktop]#

The system acts as device /sda was post installed with packages and either /dev/sda or /dev/sdb will mount the btrfs pair. Only /dev/sda will boot the pair.

Each device was initially loaded with post install packages in btrfs and now contain only /dev/sda packages. AFAICT it is not a raid system but just a jbod.
Perhaps I misread the data but 2 devices are defined and can be separately mounted as the btrfs system.

With filesystem display, the usual listing of btrfs Volumes does not list a volume for /dev/sdb, indicating that it is not a separate system item. This confirms there is but one btrfs system of two devices.
EDIT:11/19/13....Booted from grub fallback with the following systemd data resulting:

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.616s (kernel) + 1.755s (userspace) = 3.372s
[root@n6re Desktop]#

This is with the two maxell devices in jbod having a free capacity of 29.5 GB. This system runs in root only with full device capacity...no partitions and no raid. Hdparm records 85-86MB/s for the drives in read speed. The system boots to xfce4 Desktop with no log-in nor passwd with a beautiful picture of a hot air balloon in a blue sky background!

EDIT: Wheras the 3+seconds in system-analyze were actual observed times to displayed Desktop, I have been unable to reproduce the phenomenon. It obviously means that some of the packages were loaded after Desktop was reached. Both keybd and mouse were available and a right click produced the terminal for systemd-analyze display. I would be grateful for some explanation for the low numbers in systemd as well as the fact that i was only 3+ secs in real time for the system to display xfce4 with background picture derived from jpeg in /root/filesystems.

EDIT: 11/20/13.....Installed graysky's anything-sync-daemon in a three device btrfs system which had a 13 second boot time. I utilized the sync demon to place /var/log in the daemon for /dev/shm, editing /etc/asd.conf after install of anything-sync demon from the AUR.

This producedan amazing reduction in both boot time and reboot time. the results for boot time:

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.176s (kernel) + 2.774s (userspace) = 3.950s
[root@n6re Desktop]#

The reboot time was ~2 seconds to POST.

I plan to install this package , actually a script, into my remaining btrfs systems.

First system change resulted in a great improvement in maxell devices as posted below:

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.227s (kernel) + 2.891s (userspace) = 4.119s
[root@n6re Desktop]#
[root@n6re Desktop]#
[root@n6re Desktop]#
[root@n6re Desktop]# btrfs fi df /
Data, RAID0: total=3.00GB, used=1.94GB

The boot time is somewhat slower due to the 266x devices:

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.036s (kernel) + 7.594s (userspace) = 8.630s
[root@n6re Desktop]#

Data: total=8.00MB, used=0.00
System, RAID1: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, RAID1: total=1.00GB, used=136.65MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]#
[root@n6re Desktop]#
[root@n6re Desktop]#
[root@n6re Desktop]#
[root@n6re Desktop]# btrfs fi show
Label: none uuid: 21e8448f-a524-4d81-8e2f-a86e0e6ae93a
Total devices 3 FS bytes used 2.07GB
devid 3 size 14.89GB used 2.01GB path /dev/sdc
devid 2 size 14.89GB used 1.01GB path /dev/sdb
devid 1 size 14.89GB used 2.02GB path /dev/sda

Btrfs v0.20-rc1-358-g194aa4a-dirty
[root@n6re Desktop]#
This system includes the asd package. A great improvement in boot time.

EDIT: Another system created with two raid0 elements made from 26GB 533x adat CF cards and one maxell 400x udma 16GB device. The asd package is installed and the result:

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.362s (kernel) + 3.849s (userspace) = 5.211s
[root@n6re Desktop]#
[root@n6re Desktop]#
[root@n6re Desktop]# btrfs fi df /
Data, RAID0: total=3.00GB, used=2.16GB
System, RAID1: total=32.00MB, used=4.00KB
Metadata, RAID1: total=256.00MB, used=145.79MB
[root@n6re Desktop]#
[root@n6re Desktop]#
[root@n6re Desktop]# btrfs fi show
Label: none uuid: 7b16e62d-87d9-49fc-a840-121e21e254c2
Total devices 3 FS bytes used 2.30GB
devid 1 size 7.47GB used 1.03GB path /dev/sdb
devid 3 size 14.89GB used 1.28GB path /dev/sdc
devid 2 size 7.59GB used 1.25GB path /dev/sda

Btrfs v0.20-rc1-358-g194aa4a-dirty
[root@n6re Desktop]#

All systems benefited from the asd package(script). These results show ~ 5 seconds boot to xfce4 desktop without passwd or login and all are btrfs systems. The best time was ~ 3 seconds.

This approach makes best use of the computer keeping the boot time short and power off short such that time is no longer lost and maniputation of desired packages is hastened. No long waits for monster hdd's.

Edit: 11/21/13...Reinstalled triple device cf btrfs system with three 16GB devices...two adata 16GB 533x and one Silicon Power 400x device. Did not install firefox nor thunderbird but use jumanji for internet and mpv for video display instead of cmplayer and Vlc.
The results of these changes give me less than 3 secs of boot time:

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.216s (kernel) + 1.611s (userspace) = 2.828s
[root@n6re Desktop]#
[root@n6re Desktop]# btrfs fi df /
Data, RAID0: total=3.00GB, used=1.57GB
Data: total=8.00MB, used=0.00
System, RAID1: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, RAID1: total=1.00GB, used=125.83MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]#
[root@n6re Desktop]#
[root@n6re Desktop]# btrfs fi show
Label: none uuid: 928d666e-f0c9-401d-9c96-77ab732208ca
Total devices 3 FS bytes used 1.69GB
devid 2 size 14.92GB used 1.01GB path /dev/sdb
devid 3 size 14.92GB used 2.01GB path /dev/sdd
devid 1 size 14.92GB used 2.02GB path /dev/sda

Btrfs v0.20-rc1-358-g194aa4a-dirty
[root@n6re Desktop]#

It may be possible to reduce boot to less than 2 secs.

EDIT: Booted to Desktop in 2.350 secs with triple 16GB 533x devices using asd.service time of 1.04 secs.
Upgraded all systems to linux3-12.1-1

EDIT: 11/24/13..........Re arranged the three 16Gb devices into ch2, ch3 and ch3S. On my mobo, there are four Sata ports arranged in two pairs of master/slave . Thus I have utilized two master ports for the btrfs raid0 and booted from master ch2. The entire system is functional with the size of installed packages. The pertinent data is reproduced below:

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.153s (kernel) + 1.435s (userspace) = 2.589s
[root@n6re Desktop]#
[root@n6re Desktop]#
[root@n6re Desktop]# btrfs fi df /
Data, RAID0: total=3.00GB, used=1.77GB
Data: total=8.00MB, used=0.00
System, RAID1: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, RAID1: total=1.00GB, used=127.00MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]#
[root@n6re Desktop]#
[root@n6re Desktop]# btrfs fi show
Label: none uuid: 928d666e-f0c9-401d-9c96-77ab732208ca
Total devices 3 FS bytes used 1.89GB
devid 2 size 14.92GB used 1.01GB path /dev/sdc
devid 3 size 14.92GB used 2.01GB path /dev/sdb
devid 1 size 14.92GB used 2.02GB path /dev/sda

Btrfs v0.20-rc1-358-g194aa4a-dirty
[root@n6re Desktop]#

The device in ch3s is silicon power 400x 16GB and the other channels have 16GB 533x adata devices.
This leaves the ch0 IDE port open for more devices so a raid10 is probable in btrfs.
I assume this present arrangement performs because the ch3S device contains no raid0 component nor any boot data.
Perhaps a raid10 would perform also from a combo of ch2boot(master) ch3(master) btrfs raid0 and the remaining devices in ch2S and Ch3S as long as there is no boot nor raid0 required in those ports.
I may try this approach soon.

EDIT: Installed a fourth device in the open slave channel and performed....btrfs device add.... Then mounted the system in /mnt/md with mount /dev/sda /mnt/md.

Performed ....btrfs balance start /mnt/md........after which i derived the following data:
[root@n6re Desktop]# systemd-analyze
Startup finished in 1.105s (kernel) + 1.339s (userspace) = 2.444s
[root@n6re Desktop]#
[root@n6re Desktop]#
[root@n6re Desktop]# btrfs fi df /
Data, RAID0: total=7.00GB, used=1.77GB
Data: total=8.00MB, used=0.00
System, RAID1: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, RAID1: total=1.00GB, used=127.05MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]#
[root@n6re Desktop]#
[root@n6re Desktop]# btrfs fi show
Label: none uuid: 928d666e-f0c9-401d-9c96-77ab732208ca
Total devices 4 FS bytes used 1.89GB
devid 4 size 14.92GB used 1.00GB path /dev/sdd
devid 3 size 14.92GB used 3.01GB path /dev/sdc
devid 2 size 14.92GB used 2.01GB path /dev/sdb
devid 1 size 14.92GB used 3.02GB path /dev/sda

Btrfs v0.20-rc1-358-g194aa4a-dirty
[root@n6re Desktop]#

Thus it seems the raid0 remains and the new device is made raid1 metadata.

The system is in use to post this edit.

EDIT: 11/24/13 7pm....Installed a fifth CF device in btrfs system in a similar manner to the previous fourth device but installed into IDE port with adapter. Am posting in that system with the performance data following:

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.133s (kernel) + 1.388s (userspace) = 2.522s
[root@n6re Desktop]#
[root@n6re Desktop]#
[root@n6re Desktop]# btrfs fi show
Label: none uuid: 928d666e-f0c9-401d-9c96-77ab732208ca
Total devices 5 FS bytes used 1.89GB
devid 1 size 14.92GB used 3.02GB path /dev/sdb
devid 4 size 14.92GB used 1.00GB path /dev/sde
devid 5 size 14.92GB used 1.00GB path /dev/sda
devid 2 size 14.92GB used 2.01GB path /dev/sdc
devid 3 size 14.92GB used 3.01GB path /dev/sdd

Btrfs v0.20-rc1-358-g194aa4a-dirty
[root@n6re Desktop]# btrfs fi df /
Data, RAID0: total=8.00GB, used=1.77GB
Data: total=8.00MB, used=0.00
System, RAID1: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, RAID1: total=1.00GB, used=127.27MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]#

The system reboots normally. I note that the df data doesn't seem to be affected by the additional device but the filesystem shows five devices.

EDIT: 11/26/13....Re-arranged the order of the devices , placing a Silicon Power device in the IDE port which revealed its stability in hdparm as 76MB/s in either of Sataadapter or IDE adapter interface. The adata units continue to show slower performance in hdparmbut are mostly ten or more slower than the Silicon Power device.
The performance data is presented below:

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.188s (kernel) + 1.270s (userspace) = 2.458s
[root@n6re Desktop]# hdparm -t /dev/sda /dev/sdb /dev/sdc /dev/sdd /dev/sde

/dev/sda:
Timing buffered disk reads: 230 MB in 3.00 seconds = 76.59 MB/sec

/dev/sdb:
Timing buffered disk reads: 186 MB in 3.02 seconds = 61.65 MB/sec

/dev/sdc:
Timing buffered disk reads: 194 MB in 3.00 seconds = 64.64 MB/sec

/dev/sdd:
Timing buffered disk reads: 200 MB in 3.03 seconds = 66.02 MB/sec

/dev/sde:
Timing buffered disk reads: 188 MB in 3.01 seconds = 62.43 MB/sec
[root@n6re Desktop]# systemd-analyze
Startup finished in 1.188s (kernel) + 1.270s (userspace) = 2.458s
[root@n6re Desktop]# btrfs fi df /
Data, RAID0: total=8.00GB, used=3.48GB
Data: total=8.00MB, used=0.00
System, RAID1: total=8.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, RAID1: total=1.00GB, used=181.54MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]#
[root@n6re Desktop]# btrfs fi show
Label: none uuid: 928d666e-f0c9-401d-9c96-77ab732208ca
Total devices 5 FS bytes used 3.66GB
devid 3 size 14.92GB used 3.01GB path /dev/sde
devid 1 size 14.92GB used 3.02GB path /dev/sdc
devid 4 size 14.92GB used 1.00GB path /dev/sdd
devid 5 size 14.92GB used 1.00GB path /dev/sdb
devid 2 size 14.92GB used 2.01GB path /dev/sda

Btrfs v0.20-rc1-358-g194aa4a-dirty
[root@n6re Desktop]#
[root@n6re Desktop]#

EDIT: 11/23/13....Performed an install to a four device CF raid0 system. Doing this with two maxell 400x 16GB devices installed via archiso as .....mkfs.btrfs -m raid0 -d raid0 /dev/sda /dev/sdc...... This pair was packaged with pacman to provide a minimal install and boot to xfce4 Desktop. It booted in less than 5 secs.
I then prepared for the addition of two 8GB adata cf cards using device add procedure. This was accomplished after powering down to insert the devices in open ports /dev/sdb and /dev/sdd. After rebooting the following procedure was employed:

#mount /dev/sda /mnt/md  (created /mnt/md)
#btrfs device add /dev/sdb /mnt/md
# btrfs balance start /mnt/md ...(waited for balance to complete)
#btrfs device add /dev/sdd /mnt/md
#btrfs balance start /mnt/md......waited for balance to complete.

The result produced the following data:

[root@n6re Desktop]# btrfs fi show
Label: none uuid: be7f0ac5-dd06-4883-bbdd-e9a00f0c4598
Total devices 4 FS bytes used 1.66GB
devid 2 size 14.89GB used 1.06GB path /dev/sdd
devid 4 size 7.59GB used 1.06GB path /dev/sdc
devid 1 size 14.89GB used 1.09GB path /dev/sda
devid 3 size 7.47GB used 1.06GB path /dev/sdb

Btrfs v0.20-rc1-358-g194aa4a-dirty
[root@n6re Desktop]# btrfs fi df /
Data, RAID0: total=4.00GB, used=1.54GB
System: total=32.00MB, used=4.00KB
Metadata, RAID0: total=256.00MB, used=124.77MB
[root@n6re Desktop]#

Initially, the maxell pair were installed in Master Sata slots w/adapters.

I surmised that I could use an arrangement wherein master A and slave B could be utilized for the boot raid0 pair. (btrfs is not mdraid) So i switched the devices in master B and slave B and rebooted successfully.

This setup has different sized elements and performs smoothly in btrfs. They are full sized elements.

One item noted in working with these CF devices....some perform very poorly in the IDE port but others do as well as the Sata interface and still others do exactly as advertised speed wise.

EDIT 1129/13....Best post-install systemd-analyze time was 2.014secs and the best base base-devel number was 1.994 secs.
EDIT: 12/1/13....Installed linux-3.12.2-1 with good results. Reducing to three devices in all systems to avoid future problems with slave devices. Best arrangement with five devices produced 2.5 second boot time to xfce4 Desktop background pix with over 100GB capacity in root only mode.

EDIT: 12/3/13.Install with 3 full sized 16GB CF maxell 400x UDMA produced the following results:

[root@n6re Desktop]# btrfs fi show
Label: none uuid: 4e491ce7-7583-4a1e-a233-59132a5ff5ec
Total devices 3 FS bytes used 3.51GB
devid 1 size 14.89GB used 5.02GB path /dev/sda
devid 2 size 14.89GB used 5.00GB path /dev/sdb
devid 3 size 14.89GB used 5.00GB path /dev/sdc

Btrfs v0.20-rc1-358-g194aa4a-dirty
[root@n6re Desktop]# btrfs fi df /
Data, RAID0: total=12.00GB, used=3.29GB
Data: total=8.00MB, used=0.00
System, RAID0: total=15.94MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, RAID0: total=3.00GB, used=224.53MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# df -h
Filesystem Size Used Avail Use% Mounted on
/dev/sda 45G 3.6G 39G 9% /
dev 1.6G 0 1.6G 0% /dev
run 1.6G 332K 1.6G 1% /run
tmpfs 1.6G 13M 1.6G 1% /dev/shm
tmpfs 1.6G 0 1.6G 0% /sys/fs/cgroup
tmpfs 1.6G 24K 1.6G 1% /tmp
[root@n6re Desktop]# systemd-analyze
Startup finished in 1.120s (kernel) + 1.707s (userspace) = 2.828s
[root@n6re Desktop]#

As shown, nearly 4GB of data which includes jumanji,k3b,kdenlive mpv,smplayer,vlc,feh,gvfs,gparted,hdparm and thunderbird. It is a fast booting system and performs nicely with jumanji as web browser. The browser can download multiple URLs with a single command and instantly select among them after the download completes. In my system the bottleneck is adsl @768KB/s so ~20 secs is needed to download 10 url's.

EDIT: 12/5/13.Installed btrfs to triple Cf cards one a kingston 16GB 800x, another a maxell 16GB 400x udma, and the third a siliconpower 400x 16GB. These performed well with the install performed with...mkfs.btrfs -m raid0 -d raid0 /dev/sda /dev/sdb /dev/sdc....

The results after post install of pacman packages are as follows:

[root@n6re Desktop]# btrfs fi show
Label: none uuid: a578fc88-19c2-4714-9863-67cbc832c61d
Total devices 3 FS bytes used 1.80GB
devid 2 size 14.92GB used 2.00GB path /dev/sdb
devid 1 size 14.89GB used 2.02GB path /dev/sda
devid 3 size 14.89GB used 2.00GB path /dev/sdc

Btrfs v0.20-rc1-358-g194aa4a-dirty
[root@n6re Desktop]#
[root@n6re Desktop]#
[root@n6re Desktop]# btrfs fi df /
Data, RAID0: total=3.00GB, used=1.68GB
Data: total=8.00MB, used=0.00
System, RAID0: total=15.94MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, RAID0: total=3.00GB, used=123.52MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]#
[root@n6re Desktop]#
[root@n6re Desktop]# systemd-analyze
Startup finished in 1.097s (kernel) + 3.627s (userspace) = 4.725s
[root@n6re Desktop]# systemd-analyze blame
3.363s asd.service
1.166s systemd-logind.service
517ms alsa-restore.service
112ms systemd-sysctl.service
110ms sys-kernel-debug.mount
103ms systemd-udev-trigger.service
96ms polkit.service
96ms kmod-static-nodes.service
83ms udisks2.service
72ms network.service
49ms sys-kernel-config.mount
44ms systemd-update-utmp.service
22ms systemd-random-seed.service
22ms dev-mqueue.mount
20ms dev-hugepages.mount
19ms systemd-user-sessions.service
18ms systemd-journal-flush.service
18ms tmp.mount
17ms systemd-remount-fs.service
13ms systemd-tmpfiles-setup-dev.service
10ms systemd-udevd.service
[root@n6re Desktop]#

The analyze info reveals I used asd.service which took over 3 seconds. The raid0 appears for all three devices, which is of interest altho it isn't mdraid.

The results of hdparm are shown below:

[root@n6re Desktop]# hdparm -t /dev/sda /dev/sdb /dev/sdc

/dev/sda:
Timing buffered disk reads: 198 MB in 3.01 seconds = 65.88 MB/sec

/dev/sdb:
Timing buffered disk reads: 232 MB in 3.02 seconds = 76.86 MB/sec

/dev/sdc:
Timing buffered disk reads: 248 MB in 3.01 seconds = 82.33 MB/sec
[root@n6re Desktop]#

The speed for the kingston as /dev/sda is good for the IDE port while the two devices in Sata port adapters show faster read speeds with the maxell showing the highest.

EDIT: 12/6/13.Upgraded linux to 3.12.3-1 on all systems with no problems. Four triple devices in btrfs, one HDD back-up, and one raid0 single mdraid usb sdhc 16gb speed 10.

EDIT: 12/9/13....Installed btrfs in three cf cards, one 16GB and two 8GB cards. The procedure used is described as follows:

Using the archiso CF card in USB slot booted into that install.....
Began the process with....mkfs.btrfs -m raid0 -d raid0 /dev/sda /dev/sdb /dev/sdc....
(this resulted in a request for a force since the units had previous btrfs data) thus:
mkfs.btrfs -f -m raid0 -d raid0 /dev/sda /dev/sdb /dev/sdc......
(this produced a statement that the array was formed.

Proceeded to install static IP and static DNS data for the enp ethernet..........
Continued with ....mount /dev/sda /mnt....(mounts the btrfs array as root)
Entered ...pacstrap /mnt base base-devel....
After the pacstrap step was completed, ran genfstab -U -p /mnt >> /mnt/etc/fstab..

Entered arch-chroot /mnt

cd /etc and modified mkinitcpio for btrfs module and no fsck in hooks.

Performed the systemd preps for hostname, timezone,locale.gen,locale-gen, LANG=en, LC=TIME=en....

Entered pacman -S grub.... received grub-1:2.00.1282.g5ae5c54-1 for my x86_64 system and permitted it to install.

Followed that with: grub-install /dev/sda...result was accepted with no error.

Now begins the problem....entered ....grub-mkconfig -o /boot/grub/grub.cfg...resulted in a list of errors below:

#found linux image: /boot/vmlinuz-linux
#found initrd image: /boot/initramfs-linux.img
error: out of memory
error:syntax error
error: incorrect command
error: syntax error
syntax error at line 160

#/boot/grub/grub.cfg.new file attached. done

Proceeded to enter mkinitcpio -p linux which was successful.

Entered ctrl-d to escape chroot and return to archiso prompt.

Entered blkid and recorded the UUID of the btrfs system.

Rebooted into btrfs /dev/sda and completed the grub welcome, 5 second prompt anly to fail to receive a prompt for post install due to failed grub data.

Rebooted and at grub prompt edited the set root=(hd0,msdos5) to set root= (hd0) and replaced the incorrect UUID with that recorded previously. Entered F10 and obtained postinstall prompt after which I installed packages with pacman.

End of description of the newest grub interaction. The previous grub version posted a statement after grub-mkconfig complaining above device map which I ignored with no consequential problems

The grub version before that (2 versions back) performed quite normally with grubmkconfig issuing the found: statements only.

The newest grub gas a weird monicker and it performs weirdly. The UUID and set rootdata it generated with my btrfs system was that of the archiso install device!

Perhaps I need to install btrfs-progs within the arch-chroot during initial install to prevent this weird performance. But I see no real connection between grub2 and the progs since I didn't need the progs in previous grub2 installs (with the previous versions of grub2). All installs with the latest version of grub2 produce the same error of using the archiso data for the boot params.

The final activity necessary is to edit the /boot/grub data in the btrfs system to establish the correct grub boot by making the grub.cfg to grub.cfg.old and the grub.cfg.new to grub.cfg following that with mkinitcpio -p linux.

More fun!!!!!

EDIT: 12/9/13.The previous example shows the double fail of both grub and the UUID system which is supposed to be sacrosanct. I plan to backtrack the grub to an earlier version because it is supposed to be a bootloader not a game changer. I note today that one post has mentioned the failing of the latest grub which has been released with the hangups and has had a big discussion in the bug forums. To be really honest about grub, I think the original version did a bang-up job and we don't need any "improvements" to a bootloader in spite of new -fangled EFI or whatever. If the new systems need a special grub, provide it as a separate item with EFI monicker.

EDIT:...12/11/13...Performed a re-install of btrfs on a triple device btrfs and used grub obtained from USB cf card to eliminate the problem with the latest grub. I used the 5043 version with no problem. I created a /mnt dir /mnt/md to mount the device and used pacman -U to install the previous grub version during the install with archiso in the chroot. This adds some complexity to the install but avoids any new problems with grub upgrades.
Happy camper! Merry Christmas to all...

EDIT: 12/12/13...The devices used in the install were 16GB adata 266x and produce the following:

[root@n6re Desktop]# hdparm -t /dev/sda /dev/sdb /dev/sdc

/dev/sda:
Timing buffered disk reads: 144 MB in 3.01 seconds = 47.90 MB/sec

/dev/sdb:
Timing buffered disk reads: 196 MB in 3.03 seconds = 64.68 MB/sec

/dev/sdc:
Timing buffered disk reads: 188 MB in 3.01 seconds = 62.45 MB/sec
[root@n6re Desktop]# btrfs fi df /
Data, RAID0: total=6.00GB, used=2.01GB
Data: total=8.00MB, used=12.00KB
System, RAID0: total=15.94MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, RAID0: total=3.00GB, used=136.27MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]#
[root@n6re Desktop]# btrfs fi show
Label: none uuid: b09c3ddb-ca68-4f28-83ac-4c7bd3eb552a
Total devices 3 FS bytes used 2.14GB
devid 3 size 14.92GB used 3.00GB path /dev/sdc
devid 2 size 14.92GB used 3.00GB path /dev/sdb
devid 1 size 14.92GB used 3.02GB path /dev/sda

Btrfs v0.20-rc1-358-g194aa4a-dirty
[root@n6re Desktop]#

The read speed shown for /dev/sda is that obtained in the IDE port and the others are Sata port results.

The boot speed is somewhat slower due to the slower 266x devices:

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.036s (kernel) + 7.594s (userspace) = 8.630s
[root@n6re Desktop]#

Performance in general is snappy and jumanji provides multiple url's for instantaneous browsing.

EDIT: 12/14/13 ....Selected Sata enhanced mode for Sata ports on machine. Added two devices to the existing three device btrfs system. The sizes are mixed, two 32GB devices marked 600x, and three 16GB devices . Total size of the full size devices is 112GB. The pertinent datas appear below:

[root@n6re Desktop]# hdparm -t /dev/sda /dev/sdb /dev/sdc /dev/sdd /dev/sde

/dev/sda:
Timing buffered disk reads: 156 MB in 3.03 seconds = 51.51 MB/sec

/dev/sdb:
Timing buffered disk reads: 248 MB in 3.02 seconds = 82.19 MB/sec

/dev/sdc:
Timing buffered disk reads: 216 MB in 3.03 seconds = 71.37 MB/sec

/dev/sdd:
Timing buffered disk reads: 232 MB in 3.02 seconds = 76.94 MB/sec

/dev/sde:
Timing buffered disk reads: 242 MB in 3.02 seconds = 80.16 MB/sec
[root@n6re Desktop]#
[root@n6re Desktop]#
[root@n6re Desktop]# btrfs fi show
Label: none uuid: a578fc88-19c2-4714-9863-67cbc832c61d
Total devices 5 FS bytes used 2.04GB
devid 4 size 29.82GB used 1.23GB path /dev/sdc
devid 1 size 14.89GB used 2.20GB path /dev/sde
devid 2 size 14.92GB used 2.20GB path /dev/sdd
devid 3 size 14.89GB used 2.20GB path /dev/sdb
devid 5 size 29.82GB used 1.20GB path /dev/sda

Btrfs v0.20-rc1-358-g194aa4a-dirty
[root@n6re Desktop]#
[root@n6re Desktop]#
[root@n6re Desktop]# btrfs fi df /
Data, RAID0: total=8.00GB, used=1.92GB
System: total=32.00MB, used=4.00KB
Metadata, RAID0: total=1.02GB, used=125.30MB
[root@n6re Desktop]#
[root@n6re Desktop]#
[root@n6re Desktop]# systemd-analyze
Startup finished in 1.266s (kernel) + 3.153s (userspace) = 4.419s
[root@n6re Desktop]#

The device in /dev/sda is a 32GB 600x device which is being restricted to the slower speed. The result in fi df is interesting.

Attempted to run balance but one device seems to have a problem which does not permit the balance to occur. The system boots fine with the five devices and performs snappy as usual.

I recently upgraded all systems to linux-3.12.4-1 with no problems, altho I ignored upgrade for grub.

Thus, this system has one pata device in /sda and four sata based devices in sata adapters for cf cards. Device /sda has an IDE adapter for its cf card.

The speed differences are plus 50% for the sata devices vs pata device.

Have to experiment with subvolumes soon.

EDIT: 12/14/13......Re-installed four adata 533x 16GB devices with enhanced sata mode on mobo using four Cf/Sata adapters.

Used grub 5043 version for install bootloader.

Results after post install indicate a much slower performance of archlinux boot with these devices which has been consistent thru all installa with those devices. However, with just the base and base-devel installed, the systemd-analyze reported 2+ seconds.

The pertinenet dat for this install which was changed to raid10 using....mkfs.btrfs -m raid10 -d raid10 /dev/sda /dev/sdb /dev/sdc /dev/sdd is produced below:

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.173s (kernel) + 13.805s (userspace) = 14.978s
[root@n6re Desktop]# btrfs fi show
Label: none uuid: d951196f-af4c-48c2-82fc-0a2c89a9b727
Total devices 4 FS bytes used 1.87GB
devid 2 size 14.92GB used 2.01GB path /dev/sdb
devid 4 size 14.92GB used 2.01GB path /dev/sdd
devid 3 size 14.92GB used 2.01GB path /dev/sdc
devid 1 size 14.92GB used 2.03GB path /dev/sda

Btrfs v0.20-rc1-358-g194aa4a-dirty
[root@n6re Desktop]# btrfs fi df /
Data, RAID10: total=2.00GB, used=1.75GB
Data: total=8.00MB, used=0.00
System, RAID10: total=16.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, RAID10: total=2.00GB, used=123.92MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]# btrfs fi balance start /
Done, had to relocate 6 out of 6 chunks
[root@n6re Desktop]# hdparm -t /dev/sda /dev/sdb /dev/sdc /dev/sdd

/dev/sda:
Timing buffered disk reads: 200 MB in 3.02 seconds = 66.23 MB/sec

/dev/sdb:
Timing buffered disk reads: 188 MB in 3.01 seconds = 62.54 MB/sec

/dev/sdc:
Timing buffered disk reads: 188 MB in 3.01 seconds = 62.44 MB/sec

/dev/sdd:
Timing buffered disk reads: 196 MB in 3.03 seconds = 64.70 MB/sec
[root@n6re Desktop]#

The speeds indicated are consistent with the adata device params but perhaps 10MB/sec slower than expected in a sata connection.
The balance command produced a normal output statement.

This is first attempt at raid10 with my system.

AFAICT I cannot use compression with my full-size devices since they are root only and grub cannot boot compressed data which would appear in the boot area. I may be wrong about this point, however.

Actually installing compression from the command line is not outlined as a single command AFAICT such that all instructions appear on one command operation.

Raid10 just in time for Christmas!!

EDIT: 12/15/13..Upgraded all to linux-3.12.5-1-x86_64.pkg.tar.xz with no problems. I am wondering what the fi df data means for system in raid10 system using four devices. My install was -m raid10 -d raid10 for four devices. I now have one five device raid0 setup which was built by adding devices to a three device system...a four device raid10 system and a three device system in raid0. All are btrfs full sized devices. with no partitions, root only. The system in use herein is the five device arrangement.

EDIT: 12/21/13.... Upgrade to four device raid10 failed to reboot which was due to a loss of pacman data repo as determined in chroot mode. This may be due to btrfs fail or to corrupt download.
I have re-installed the devices to two paired btrfs raid0 systems which perform well with the following data:

[root@n6re Desktop]# hdparm -t /dev/sda /dev/sdb

/dev/sda:
Timing buffered disk reads: 188 MB in 3.00 seconds = 62.58 MB/sec

/dev/sdb:
Timing buffered disk reads: 188 MB in 3.01 seconds = 62.37 MB/sec
[root@n6re Desktop]#
[root@n6re Desktop]# btrfs fi show
Label: none uuid: dd28c58e-ff34-4cdd-8552-2bfbce60ff5e
Total devices 2 FS bytes used 2.34GB
devid 1 size 14.92GB used 3.03GB path /dev/sda
devid 2 size 14.92GB used 3.01GB path /dev/sdb

Btrfs v0.20-rc1-358-g194aa4a-dirty
[root@n6re Desktop]#
[root@n6re Desktop]# btrfs fi df /
Data, RAID0: total=4.00GB, used=2.21GB
Data: total=8.00MB, used=6.19MB
System, RAID0: total=16.00MB, used=4.00KB
System: total=4.00MB, used=0.00
Metadata, RAID0: total=2.00GB, used=127.85MB
Metadata: total=8.00MB, used=0.00
[root@n6re Desktop]#
This is typical for the two btrfs systems which boot in 6+ seconds.

I also installed btrfs to two 8GB adata 266x devices with success.

Additionally, I reprogrammed a 16GB sdhc 10 speed mdraid single device raid0 to be a single btrfs system in archlinux. It boots in analyze spped of 7+ seconds but display of Desktop is delayed to about 20 seconds. I may spend some research time to determine what may cause this. The hdparm speed is 17MB/s.

EDIT: Decided to break up the five device system in favor of a pair of 32GB silicon power devices in btrfs raid0. The install was made and the results a given below:

root@n6re Desktop]# systemd-analyze
Startup finished in 1.168s (kernel) + 1.592s (userspace) = 2.760s
[root@n6re Desktop]#
[root@n6re Desktop]# btrfs fi df /
Data, RAID0: total=2.00GiB, used=1.80GiB
Data, single: total=8.00MiB, used=0.00
System, RAID0: total=16.00MiB, used=4.00KiB
System, single: total=4.00MiB, used=0.00
Metadata, RAID0: total=2.00GiB, used=126.56MiB
Metadata, single: total=8.00MiB, used=0.00
[root@n6re Desktop]#
[root@n6re Desktop]# btrfs fi show
Label: none uuid: 0a994893-accf-48f0-81ac-cc8ff009f064
Total devices 2 FS bytes used 1.92GiB
devid 1 size 29.82GiB used 2.03GiB path /dev/sdb
devid 2 size 29.82GiB used 2.01GiB path /dev/sdc

Btrfs v3.12
[root@n6re Desktop]# hdparm -t /dev/sdb /dev/sdc

/dev/sdb:
Timing buffered disk reads: 212 MB in 3.02 seconds = 70.28 MB/sec

/dev/sdc:
Timing buffered disk reads: 214 MB in 3.01 seconds = 71.03 MB/sec
[root@n6re Desktop]#

The system boot time is especially fast and may be due to the equal sized higher -speed silicon power 32GB devices. This shows the difficulty in predicting the performance of devices since these devices are most likely MLC and that should slow things down. Altho, the actual install size is no where near the max and may be the reason for the speedy boot.

The boot time in systemd-analyze is very close to the actual time from grub prompt to Desktop display of xfce4 background pix. This pix takes a few extra ms to complete so the actual time seen by the user is a little more than the 2.760 seconds AFAICT. This is extremely fast imhop.

I have firefox,thunderbird mpv flashplugin and jumanji installed and running.

EDIT: 12/27/13....Upgraded all systems to linux-3.12.6-1 with no problems.

EDIT: 12/30/13...Installed kdenlive (520mb installed) into 64GB btrfs pair raid0 and obtaine 4.356sec boot time and 2 second poweroff.
Researched zfs filesystem on google and archlinux wiki. Decided that the system is not native linux, requiring special attention when upgrading kernels and subject to non-compliant changes in the future...such that I won't pursue that filesystem at this time. In fact, btrfs provides a linux based system with excellent performance in my view in spite of negative remarks from zfs practitioners.
One factor in this decision to skip zfs is the lack of forum posts on the filesystem in archlinux forums.
Addition this date of kdenlive with 520 mb of install size produced very slight change in boot time. In fact, one boot time was as low as 3.864 sec!
I hope in the near future to try a pair of 64GB compact Flash devices 600x to compare the performance. Experiments reported herein suggest that larger capacity devices perform better in btrfs systems. The present devices are Silicon Power 32GB 600x and perform as reported above.
As the boot time is attained with no tweaking of params, it seems even faster times can be attained...be reminded that the time reported in these experiments are from grub prompt to xfce4 Desktop with background picture displayed. There is no menu nor icons displayed, these are provided by mouse actions.

When installing these systems, the systemd-analyze report at the root prompt post install are typically 1.9 seconds.

Happy New Year!!!!

EDIT: 1/2/14.....Installed btrfs to two maxell 16GB 400x udma compact flash devices.

Install command was ....mkfs.btrfs -m raid0 -d raid0 /dev/sda /dev/sdb.....

The usual install result with btrfs was a three raid0 arrangement as previously posted.

Then after post install of packages, provided additional maxell 16gb devices in /dev/sdc and /dev/sdd.

Booted to btrfs and mounted the btrfs system via /dev/sda in /mnt/md.

Performed ...btrfs device add /dev/sdc /mnt/md
Performed ....btrfs device add /dev/sdd /mnt/md
Performed ....btrfs balance start /mnt/md.....
The results are given below:

[root@n6re Desktop]# btrfs fi show
Label: none uuid: 09b1b4a9-306e-4baf-a42f-8e1f0d60427b
Total devices 4 FS bytes used 2.02GiB
devid 1 size 14.89GiB used 2.25GiB path /dev/sda
devid 2 size 14.89GiB used 2.25GiB path /dev/sdc
devid 3 size 14.89GiB used 2.25GiB path /dev/sdb
devid 4 size 14.89GiB used 2.28GiB path /dev/sdd

Btrfs v3.12
[root@n6re Desktop]# btrfs fi df /
Data, RAID0: total=8.00GiB, used=1.90GiB
System, single: total=32.00MiB, used=4.00KiB
Metadata, RAID0: total=1.00GiB, used=122.29MiB
[root@n6re Desktop]# hdparm -t /dev/sda /dev/sdb /dev/sdc /dev/sdd
bash: hdparm: command not found
[root@n6re Desktop]# pacman -S hdparm
resolving dependencies...
looking for inter-conflicts...

Packages (1): hdparm-9.43-2

Total Download Size: 0.07 MiB
Total Installed Size: 0.19 MiB

:: Proceed with installation? [Y/n]
:: Retrieving packages ...
hdparm-9.43-2-x86_64 68.5 KiB 108K/s 00:01 [######################] 100%
(1/1) checking keys in keyring [######################] 100%
(1/1) checking package integrity [######################] 100%
(1/1) loading package files [######################] 100%
(1/1) checking for file conflicts [######################] 100%
(1/1) checking available disk space [######################] 100%
(1/1) installing hdparm [######################] 100%
Optional dependencies for hdparm
bash: for wiper.sh script [installed]
[root@n6re Desktop]# hdparm -t /dev/sda /dev/sdb /dev/sdc /dev/sdd

/dev/sda:
Timing buffered disk reads: 250 MB in 3.02 seconds = 82.73 MB/sec

/dev/sdb:
Timing buffered disk reads: 194 MB in 3.01 seconds = 64.41 MB/sec

/dev/sdc:
Timing buffered disk reads: 248 MB in 3.01 seconds = 82.50 MB/sec

/dev/sdd:
Timing buffered disk reads: 244 MB in 3.01 seconds = 81.11 MB/sec
[root@n6re Desktop]#
I note the speed of /dev/sdb is awry and it suggests that that device may be dying or the adapter is defective. It may be the single device referred to in the filesystem list df.

The combo is performing properly in archlinux and exists only experimentally. It did not produce raid10 for four device btrfs system.

I am wondering if one can convert btrfs systems from raid0 to raid 10 with the convert command. May have a shot at that before long.

I am thankful for the interest shown in my post since it is now over 10,000 views and edits.

My use of btrfs commands may be unique, setting up odd combos in btrfs filesystems.

EDIT:1/3/14...After several reboots, the /dev/sdb read speed was stabilized at 86MB/s the same as the rest.

EDIT: Performed chattr +c on several directories in a btrfs system. Verified their compression with lsattr but do not expect much to change on my smallish systems. It was done to learn about the function of compression and the commands required. Have not done and subvolume activity.
EDIT: 1/11/14...Performed....chattr -R +c lz4 /.... to install compression to two separate btrfs systems. This command took some time to complete an d the result was displayed by the command...lsattr /.... which shows a "c" in each directory or file affected by the compression algorithm and reports all those incompatible with compression. The result is shown below;

[root@n6re Desktop]# lsattr /
--------c------- /var
lsattr: Inappropriate ioctl for device While reading flags on /dev
lsattr: Inappropriate ioctl for device While reading flags on /run
--------c------- /etc
lsattr: Inappropriate ioctl for device While reading flags on /tmp
lsattr: Inappropriate ioctl for device While reading flags on /sys
lsattr: Inappropriate ioctl for device While reading flags on /proc
--------c------- /usr
lsattr: Operation not supported While reading flags on /bin
--------c------- /boot
--------c------- /home
lsattr: Operation not supported While reading flags on /lib
lsattr: Operation not supported While reading flags on /lib64
--------c------- /mnt
--------c------- /root
lsattr: Operation not supported While reading flags on /sbin
--------c------- /srv
--------c------- /Mom and Dad.jpg
--------c------- /P1020003.JPG
--------c------- /jumanji
--------c------- /jumanji1
--------c------- /jumanji3
--------c------- /Family pics 1.jpg
--------c------- /geese.png
--------c------- /opt
[root@n6re Desktop]#
I am not expecting much change in performance with this install of lz4 but the mechanics of the operation are instructive. The lz4 algorithm is the latest to be applied in Linux systems and is supported in the latest kernel and mkinitcpio.

The system herein utilized has the lz4 algorithm installed in compact flash devices as shown above.

The command utilized seems to be universal and is applied to a running btrfs system, in my case, a root -only system.

EDIT: 1/12/14 .... Upgraded all systems to linux-3.12.7-1-x86_64

EDIT: 1/13/14......Utilized a single device btrfs....adata 8GB 266x CF card...to enter a subvolume. The command ...btrfs subvolume create video...created the subvolume in /root/Desktop with the ID video. Then ran ...btrfs subvolume list -p . ..... this listed video as a subvolume.

I am wondering if I can call up a second btrfs volume, ....such as...btrfs subvolume create /dev/sdx..... wherein /dev/sdx is another complete raid0 pair. I recognize the existence of btrfs volumes in the filesystem menu which are available via gvfs and they can be mounted. Curious to know what happens... if the volumes will act differently ...when made subvolumes. This requires enough space be provided in the parent btrfs volume no doubt.
Perhaps the use of the filesystem menu with gvfs is enough for all purposes since it allows mounting an entire btrfs pair for use as needed for data stored therein.

EDIT:1/14/14.....Created a subvolume for a 32GB btrfs raid0 pair with an 8GB btrfs volume known as /sdc with....btrfs subvolume create /sdc....

Ran the following command to verify the subvolume:

[root@n6re Desktop]# btrfs subvolume show /sdc
/sdc
Name: sdc
uuid: 788527a9-a5f8-ca4f-8878-37e767a42995
Parent uuid: -
Creation time: 2014-01-14 09:23:37
Object ID: 260
Generation (Gen): 4314
Gen at creation: 4314
Parent: 5
Top Level: 5
Flags: -
Snapshot(s):
[root@n6re Desktop

This verifies that the subvolume was created with an ID=260 and UUID as shown.

Perhaps this is a useable arrangement not yet obvious to this user.
EDIT: 1/14/14...Googling revealed that the subvolume is not a block device so the trial performed via /dev/sdc is not a seletion of the volume on /dev/sdc but represents a subvolume generated with the name sdc and has a uuid assigned to it. Thus, block devices are not capable of being subvolumes.
The volume provided by other btrfs devices is signified as such in the menu per gvfs with the size given and can be mounted.

Use of the subvolume is not fully understood by this user.

EDIT: 1/15/14...Upgrade to linux-3.12.7-2-x86....on all systems with no problems.

EDIT: 1/16/14.Made a decision to break up the 32GB btrfs raid0 pair of Silicon Power devices and made them single btrfs devices. Installed them with the latest beta grub with no problem. The results are provided below with the first systemd analysis following the post install for each device done separately was 3.065 secs for one unit and 2.967 secs for the second device. These values increased to approximately 3.5 secs after several reboots. The data for one of the 32GB devices is shown below;

[root@n6re Desktop]# btrfs fi show
Label: none uuid: d518c16c-59b5-44d8-a59c-4fab409e73e9
Total devices 1 FS bytes used 2.64GiB
devid 1 size 29.82GiB used 5.04GiB path /dev/sdb

Btrfs v3.12
[root@n6re Desktop]# btrfs fi df /
Data, single: total=3.01GiB, used=2.50GiB
System, DUP: total=8.00MiB, used=16.00KiB
System, single: total=4.00MiB, used=0.00
Metadata, DUP: total=1.00GiB, used=143.58MiB
Metadata, single: total=8.00MiB, used=0.00

[root@n6re Desktop]# hdparm -t /dev/sdb

/dev/sdb:
Timing buffered disk reads: 222 MB in 3.00 seconds = 73.98 MB/sec
[root@n6re Desktop]# hdparm -I /dev/sdb

/dev/sdb:

CompactFlash ATA device
Model Number: SILICON POWER
Serial Number: 00010896
Firmware Revision: 20110620
Standards:
Likely used: 6
Configuration:
Logical max current
cylinders 62041 62041
heads 16 16
sectors/track 63 63
--
CHS current addressable sectors: 62537328
LBA user addressable sectors: 62537328
Logical/Physical Sector size: 512 bytes
device size with M = 1024*1024: 30535 MBytes
device size with M = 1000*1000: 32019 MBytes (32 GB)
cache/buffer size = 1 KBytes (type=DualPort)
Capabilities:
LBA, IORDY(can be disabled)
bytes avail on r/w long: 4
Standby timer values: spec'd by Vendor
R/W multiple sector transfer: Max = 1 Current = 0
Advanced power management level: disabled
DMA: mdma0 mdma1 mdma2 udma0 udma1 udma2 udma3 udma4 *udma5 udma6
Cycle time: min=120ns recommended=120ns
PIO: pio0 pio1 pio2 pio3 pio4
Cycle time: no flow control=120ns IORDY flow control=120ns
Commands/features:
Enabled Supported:
Power Management feature set
Write cache
WRITE_BUFFER command
READ_BUFFER command
NOP cmd
CFA feature set
Advanced Power Management feature set
Mandatory FLUSH_CACHE
* CFA advanced modes: pio5 pio6 mdma3 mdma4
* CFA Power Level 1 (max 500mA)
HW reset results:
CBLID- above Vih
Device num = 0
Integrity word not set (found 0x0000, expected 0xa4a5)
[root@n6re Desktop]#

The devices are marked as 600x but hdparm ID's them as udma5 and the speed shown is compatible with that level.

This performance level seems to confirm that larger devices are better altho I would reserve that judgement until I have tried a 64GB device.

The performance in boot and shutdown of the single units is desired and raid0 is thus avoided.

As a side note, maxell devices that I have utilized for years in mdadm raid arrays do not perform well with btrfs . So I have disbanded the four device raid0 array of those devices. Programming one of them as a single produced around 10 secs boot time. This in spite of the speeds of 86MB/s.

I hope to try a 64GB device asap, probably a Silicon Power device.

Then maybe try a 1000 mode cf card!

I remind all that the boot time is to Desktop in xfce4!

EDIT: 1/17/14...Installed btrfs into single 16GB mini usb device and installed xfce4 , thunderbird, opera and after establishing systemd network and xfce4, thunderbird and opera, ran .....chattr -R +c lz4 /..... This provided the following result in the mini usb device install:

[root@n6re Desktop]# lsattr /
--------c------- /var
lsattr: Inappropriate ioctl for device While reading flags on /dev
lsattr: Inappropriate ioctl for device While reading flags on /run
--------c------- /etc
lsattr: Inappropriate ioctl for device While reading flags on /tmp
lsattr: Inappropriate ioctl for device While reading flags on /sys
lsattr: Inappropriate ioctl for device While reading flags on /proc
--------c------- /usr
lsattr: Operation not supported While reading flags on /bin
--------c------- /boot
--------c------- /home
lsattr: Operation not supported While reading flags on /lib
lsattr: Operation not supported While reading flags on /lib64
--------c------- /mnt
--------c------- /opt
--------c------- /root
lsattr: Operation not supported While reading flags on /sbin
--------c------- /srv
--------c------- /jumanji
--------c------- /jumanji1
--------c------- /jumanji3
--------c------- /Mom and Dad.jpg
[root@n6re Desktop]#
Thus, the major elements of the install have compression applied and the expectataion is that any new files added will be compressed as well.
File system data appears below:

[root@n6re Desktop]# btrfs fi df /
Data, single: total=2.01GiB, used=1.66GiB
System, DUP: total=8.00MiB, used=16.00KiB
System, single: total=4.00MiB, used=0.00
Metadata, DUP: total=1.00GiB, used=96.97MiB
Metadata, single: total=8.00MiB, used=0.00
[root@n6re Desktop]# btrfs fi show
Label: none uuid: 79146b12-aa59-46f1-a302-d9e93d86b14f
Total devices 1 FS bytes used 1.76GiB
devid 1 size 14.91GiB used 4.04GiB path /dev/sdg

Btrfs v3.12
[root@n6re Desktop]# systemd-analyze
Startup finished in 2.953s (kernel) + 2.400s (userspace) = 5.353s
[root@n6re Desktop]#

Hdaprm shows:

[root@n6re Desktop]# hdparm -tT /dev/sdg

/dev/sdg:
Timing cached reads: 1960 MB in 2.00 seconds = 979.61 MB/sec
Timing buffered disk reads: 64 MB in 3.02 seconds = 21.16 MB/sec
[root@n6re Desktop]#

This data is provided by the mini drive during this post. It takes 20 seconds to reach Desktop with the mini drive.

EDIT: 1/22/14 Up graded all systems to linux-3.12.8-1-x86_64with no problems. Mymobo limits the IDE speed to 100mhz and that limits udma to mode 5. I would need to go to a new mobo to utilize the latest mode 7 devices and it isn't worth the money involved to accomplish very little. At udma5 level, the boot times are mostly 5 seconds to 7 seconds and quite acceptable. Single device btrfs provides good performance so I have changed all to singles. I have firefox in some and opera in others...thunderbird in all. Jumanji is also provided in all systems. I hope for f3fs to be bootable in grub soon similar to the arrangement with btrfs/ Perhaps the beta release will have it?

EDIT: 1/26/14....Upgraded linux-3.12.9-1 to all systems. Also included a btrfs micro san disk usb 16GB device which gave a systemd-analyze result of 18+ seconds in usb2.0 slot. Dont't know the device rating for that 16gb san disk.

EDIT: 1/28/14....Knowing that the system is relegated to udma5, I surmised I might get a benefit from faster cells in more modern compact flash devices. Therefore, i purchased a pro spec 16GB device with udma7 capability. I installed btrfs and do indeed detect a much faster handling of video files, a 229MB file transferred in ms. This may be due to cow or may be cached due to the nature of the record. I will investigate the possibilities with further study. It seems to make sense that faster cells would not slow down whether the system is IDE udma 7 capable or not.

Data obtained for this device appears below:

root@n6re Desktop]# btrfs fi df /
Data, single: total=4.01GiB, used=3.79GiB
System, DUP: total=8.00MiB, used=16.00KiB
System, single: total=4.00MiB, used=0.00
Metadata, DUP: total=1.00GiB, used=151.16MiB
Metadata, single: total=8.00MiB, used=0.00
[root@n6re Desktop]# btrfs fi show
Label: none uuid: 93ec1ba1-f9c2-4ff8-bb8e-6de0b67f77cd
Total devices 1 FS bytes used 3.94GiB
devid 1 size 14.92GiB used 6.04GiB path /dev/sda

Btrfs v3.12
[root@n6re Desktop]# systemd-analyze
Startup finished in 1.173s (kernel) + 1.951s (userspace) = 3.125s
[root@n6re Desktop]#

The systemd time is especially remarkable for a GUI boot to xfce4. This system is booted into full-sized compact flash w/o partitioning using grub and operates as root only. There are no subvolumes in this install.

EDIT: 1/30/14 Upgraded linux-3.12.9-2 into all btrfs systems (8) with no problems encountered.

EDIT: 1/31/14.....Booted to xfce4 Desktop in 16GB btrfs single device in 2.652 secs, device is ProSpec udma 7. Rebooted one hour later in 3.095 secs.

EDIT: 2/1/14....Installed otter-browser from AUR and set up bookmarks across the top panel. After reaching Desktop, 4 seconds needed for automatic display of selected bookmark from last boot-up. At present, 8 URL's provided and each instantly selectable. Otter-browser is new and in flux (one month old) and will change as time goes by but is useable at this present level.

EDIT: 2/14/14...Have otter-browser and dwb installed in btrfs device prospec 16GB udma7 and can use them simultaneously in alternate workspaces or in same workspace. Upgrade daily to five btrfs devices. Not all devices boot as quickly as the udma7 device since each device has different speed limits and internal architechure..algorithms.. Shutdown occurs in 2 seconds.

EDIT: 2/19/14....Booted btrfs 16GB device in less than 3 seconds to xfce4 Desktop...........

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.150s (kernel) + 1.821s (userspace) = 2.971s
[root@n6re Desktop]#

This seems to occur daily upon first turn-on.

EDIT: 2/20/14......Upgraded all 8 systems to linux-3.13.4-1. Added openshot package to ProSpec 16GB udma7 device. Now has kdenlive and openshot. Running otter-browser and jumanji and can run both at same time.

Additional info...Removed the files in /var/log/journal and changed /etc/systemd/journald.conf item .... SystemdMaxuse=50M....
This action produced a repeatable 2.5 second boot time in this ProSpec device. Additionally, the saved otter-browser and jumanji items were directly displayed upon reaching Desktop.

EDIT: 2/22/14.......Upon boot up this date the following data was obtained on the ProSpec single 16GB device (and all other devices installed as volumes):

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.244s (kernel) + 970ms (userspace) = 2.214s
[root@n6re Desktop]# btrfs fi show
Label: none uuid: 93ec1ba1-f9c2-4ff8-bb8e-6de0b67f77cd
Total devices 1 FS bytes used 5.84GiB
devid 1 size 14.92GiB used 10.04GiB path /dev/sdb

Label: none uuid: d518c16c-59b5-44d8-a59c-4fab409e73e9
Total devices 1 FS bytes used 4.38GiB
devid 1 size 29.82GiB used 7.04GiB path /dev/sde

Label: none uuid: 0dc6663c-e090-4585-9dc2-638c7cd599ab
Total devices 1 FS bytes used 2.16GiB
devid 1 size 14.92GiB used 5.04GiB path /dev/sdd

Label: none uuid: 036510c8-20c4-4a3e-9497-9ac01aab9588
Total devices 1 FS bytes used 2.42GiB
devid 1 size 29.82GiB used 5.04GiB path /dev/sdc

Label: none uuid: 4fbaf526-b7c2-47e5-9e02-09f465897a01
Total devices 1 FS bytes used 3.55GiB
devid 1 size 14.92GiB used 7.04GiB path /dev/sda

Btrfs v3.12
[root@n6re Desktop]# btrfs fi df /
Data, single: total=8.01GiB, used=5.67GiB
System, DUP: total=8.00MiB, used=16.00KiB
System, single: total=4.00MiB, used=0.00
Metadata, DUP: total=1.00GiB, used=174.33MiB
Metadata, single: total=8.00MiB, used=0.00

[root@n6re Desktop]# hdparm -tT /dev/sdb

/dev/sdb:
Timing cached reads: 2022 MB in 2.00 seconds = 1011.31 MB/sec
Timing buffered disk reads: 252 MB in 3.01 seconds = 83.83 MB/sec
[root@n6re Desktop]#

The boot time indicates less than a second to boot the userspace and my system is limited to udma5 with the mobo I have. Because the ProSpec is connected via an adapter to SATA ports, I seem to obtain faster performance although it may be errors in hdparm.

The best boot time I have seen is 2.151 with this ProSpec 16GB device. This performance may be due to faster chips in the device since it is udma7 capable. Mobo cannot handle that mode in IDE related ports.

Additional data shows the systemd-analyze blame:

Startup finished in 1.244s (kernel) + 970ms (userspace) = 2.214s
[root@n6re Desktop]# systemd-analyze blame
207ms systemd-logind.service
172ms systemd-udev-trigger.service
172ms alsa-restore.service
134ms systemd-sysctl.service
133ms sys-kernel-debug.mount
127ms polkit.service
122ms kmod-static-nodes.service
105ms systemd-journal-flush.service
103ms udisks2.service
68ms network.service
43ms systemd-user-sessions.service
33ms sys-kernel-config.mount
32ms dev-hugepages.mount
28ms systemd-update-utmp.service
27ms systemd-random-seed.service
26ms tmp.mount
26ms systemd-remount-fs.service
23ms systemd-tmpfiles-setup-dev.service
19ms systemd-tmpfiles-clean.service
15ms dev-mqueue.mount
9ms systemd-udevd.service
[root@n6re Desktop]#

I will stop alsa-restore.service and see the effect.....

Cannot stop alsa-restore.service...... ran the prospec device in alternate boot mode and received the following result:

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.145s (kernel) + 1.006s (userspace) = 2.152s

[root@n6re Desktop]# journalctl --disk-usage
Journals take up 11.0M on disk.
[root@n6re Desktop]#

It seems likely I can reach under 2 secs because I have had less than 1.000
secs userspace at times.

Obtained the following data when utilizing the alternate boot in grub:

[root@n6re Desktop]# systemd-analyze blame
181ms udisks2.service
180ms systemd-logind.service
170ms systemd-udev-trigger.service
149ms systemd-sysctl.service
146ms sys-kernel-debug.mount
134ms kmod-static-nodes.service
126ms polkit.service
71ms alsa-restore.service
45ms network.service
27ms systemd-user-sessions.service
23ms systemd-random-seed.service
22ms systemd-update-utmp.service
21ms systemd-remount-fs.service
21ms dev-mqueue.mount
20ms sys-kernel-config.mount
17ms dev-hugepages.mount
17ms systemd-tmpfiles-setup-dev.service
15ms systemd-udevd.service
12ms tmp.mount
9ms systemd-journal-flush.service
[root@n6re Desktop]#

The alsa-restore.service is now 71ms and the boot time is below:

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.153s (kernel) + 1.033s (userspace) = 2.187s

EDIT: 2/24/14.....Upgraded all systems to linux-3.13.5-1 x86_64 with no problems. Installed btrfs to 16GB micro sdhc card from walmart and it shows 5.11 seconds in systemd-analyze in linux-3.13.5-1. The display time to Desktop is about 10 seconds more. Activity in that device is not slow AFAICT.. I do not know the speed rating for that device. It is connected direct to a usb2 cable.

EDIT: The latest kernel now increases the boot time as shown from the system-analyze data for the prospec device:
[root@n6re Desktop]# systemd-analyze

Startup finished in 1.390s (kernel) + 1.095s (userspace) = 2.486s
[root@n6re Desktop]#

EDIT: 2/26/14 ....With journal log deleted, the following boot time resulted with the prospec device:
[root@n6re Desktop]# systemd-analyze

Startup finished in 1.069s (kernel) + 910ms (userspace) = 1.980s
[root@n6re Desktop]#

The lowest kernel reading so far....

EDIT: Several examples of lower readings:

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.069s (kernel) + 910ms (userspace) = 1.980s
[root@n6re Desktop]#

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.088s (kernel) + 802ms (userspace) = 1.891s
[root@n6re Desktop]#

In fallback mode:

[root@n6re Desktop]# systemd-analyze
Startup finished in 1.106s (kernel) + 786ms (userspace) = 1.892s
[root@n6re Desktop]#

These are typical now with the prospec unit. Fallback mode reduces the time to ~50% of the value in normal boot.

EDIT: 2/28/14...
.

Booted from16GB micro usb sdhc with the following result: (single device)

[root@n6re Desktop]# systemd-analyze
Startup finished in 2.991s (kernel) + 2.424s (userspace) = 5.416s
[root@n6re Desktop]#

The boot time for the prospec pair has stabilized at ~1.892 secs.

I have not tried a custom kernel....yet.

EDIT: 3/3/14...........Find that the adata 400x udma devices fail to respond to dma which results in failed btrfs use as bootable devices. However, if not used as bootable, these devices provide btrfs volumes, full-capacity if just established as btrfs devices. They then appear as useable volumes in a btrfs system boot to xfce4. Much data can be entered therein and accessed by the system.

Furthermore,raid0 can be applied to a pair of adata 400x udma devices, which I have done. This permitted a transfer rate of 220MB/s from the raid0 adata pair to /var/local of the mounted raid0 pair of prospec udma7 devices in the operating system.
Furthermore, using a single adata device as a btrfs volume in USB permits a 30MB/s transfer rate using a USB 3.0 adapter (USROBOTICS) . This adapter is connected to a USB 2.0 port. with a 1 meter cable.

It is likely that my mobo cannot exceed ~225MB/s due to bandwidth limits. (945gcm S2 @800mhz)

The arrangement thusly provided uses four of my ports with two inactive until accessed. The boot time in the operating system pair is just under 2 seconds with the setup described and is not affected by the other pair used as btrfs volumes (mountable via run/media/root with a mouse click.).
I must run this arrangement now for a testing period and perhaps a way to improve the operating system performance.

EDIT: 3/6/14....Experimented with USB devices using USROBOTICS USB3.0 adapters (2) on USB2.0 ports.
Installed Adata 16GB cF cards 533x in each usb adapter. Ran mkfs.btrfs -m raid0 -d raid0 /dev/sde /dev/sdj ... received error message to use -f, re-ran mkfs.btrfs with -f option. Proceeded to install 1GB video file in /dev/sde(Now the ID of the raid0 usb array).
Performed ....mount /dev/sde /mnt/md.......
Performed .....btrfs fi balance start /mnt/md.....

This resulted in the balancing of the usb raid0 array.

This array appears in xfce4 Desktop filesystem menu as two 16GB volumes, one of them is mountable and can be interroated with a mouse-over before selection to mount.

Btrfs fi show provides a listing of the usb devices as a btrfs paired array.

A test of transfer speed of the 1GB file to /tmp showed a transfer speed of 46MB/s. Hdparm shows a read sopeed of 23MB/s on both usb devices.

My mobo does not have usb3.0 capability.

The setup with usbrobotics allows 5 device slots on each adapter . This allows ten additional device listings in xfce4 each mountable separately or arranged in a raid array as has been demonstrated.
This arrangement makes it possible to generate a raid0 array directly in archlinux OS.
One application would be to copy /var/cache/pacman/pkg into the array for example. Packages obtained from AUR could also be copied into the array.
Probably the array could be used to run PKGBUILDS.
Perhaps it may also be used to provide storage for downloads.
Since these compact Flash devices are useable in IDE config, the raid0 pair could run in the sata adapters for higher speed performance.
The devices can be removed and reinserted without loss of data.
With all usb slots filled with devices, the operating system speeds for power off is longer and the display time in boot-up is affected. This does not occur with the raid0 array only being inserted. The boot time is ~2 secs and the poweroff time 2 secs in that event.

This experiment demonstrates the versatility of the btrfs system in archlinux. All experiments were conducted on an operating archlinux system boot-up.

Uo to this time, reboots have been performed many times without failure of the installed raid0 system. I have yet to check it during and after a kernel upgrade.

Because these devices are usb, they can be removed at will. This cannot be done (without ahci) in IDE interfaces. Multiple raid pairs are piossible as well by inserting them in the adapter at will.

Would appreciate a usb 3.0 system!!!! Maybe can do that with pci-e?

EDIT: 3/7/14....Installed a dual 3.0 pcie adapter and utilized the compact flash 16GB 533x adata cards in a raid0 config. The hdparm speed reported 33MB/s for each drive but I do not know if hdparm results apply to 3.0 device connections. The boot time is still ~2 seconds and the ability to access multiple sdhc and compact flash devices is fabulous. The pcie cannot be used to boot AFAICT (not allowed in bios). The result of applying a pcie device is to free up two usb 2.0 ports for even more access to storage. The POST time is unaffected and the devices used in the usb adapters (us robotics) are removable.
Transfer tests of the raid0 using timed copy shows 50MB/s with my mobo which has 100mhz bus for pcie. Thus, bandwidth is limited.

All ports of the pcie adapters are displayed and are accessable in xfce4 menu and can be read and written to in the booted arch system.

Btrfs allows setting up the raid0 system in the pcie while operating in archlinux system. Thus additional raid devices can be set-up at will and inserted into the adapters in the OS while operating in archinuxOS.

I hesitate to move up to new mobo due to efi modes required which might be destructive to my fast boot system.

I use compact flash devices because they are more versatile than SSD drives IMHOP, since they are removable at will. With adapters in sata ports they are also bootable in 2 seconds (and less).

Happy camper!!

EDIT: 3/8/14...The following data applies to the USB 3.0 pair of compact flash devices installed in my flash-based arcglinux.....

sh-4.2# btrfs fi show
Label: none uuid: fdf5f2d3-b078-41ab-a56c-88ea58e0e91b
Total devices 2 FS bytes used 3.65GiB
devid 1 size 14.92GiB used 4.03GiB path /dev/sda
devid 2 size 14.92GiB used 4.01GiB path /dev/sdc

Label: none uuid: ca993e36-9d97-490b-9b8a-526e4fe2860d
Total devices 2 FS bytes used 1.83GiB
devid 1 size 14.92GiB used 1.28GiB path /dev/sdd
devid 2 size 14.92GiB used 1.28GiB path /dev/sdi

Label: none uuid: 4fbaf526-b7c2-47e5-9e02-09f465897a01
Total devices 1 FS bytes used 3.70GiB
devid 1 size 14.92GiB used 7.04GiB path /dev/sdb

Btrfs v3.12
sh-4.2# mount /dev/sdi /mnt/md
sh-4.2# btrfs fi df /mnt/md
Data, RAID0: total=2.00GiB, used=1.83GiB
System, RAID1: total=32.00MiB, used=16.00KiB
Metadata, RAID1: total=256.00MiB, used=2.17MiB
sh-4.2#

The technique seems to be stable and re-bootable. The hdparm readings which may be erroneous show 33MB/s read speed.
The pair were generated with...btrfs -d raid0 /dev/sdd /dev//sdi and as is shown, /dev/sdi was used to mount the pair in raid mode in /mnt/md. Since the btrfs command cannot be used for details such as ....fi df /... the target is merely changed to the /mnt/md condition of the raid pair for data retrieval.

There is 2GB of data stored in that array at this writing. It is usual to add data while mounted in /mnt/md.

EDIT: ...Opened the raid0 into /run/media/root (the normal mount for the data files exp\ternel to the system and selected three files from the data therein. The total size of those files is 1640MB. The limit of /tmp is 1.6GB, somewhat larger than the data selected.....results below:
sh-4.2# time cp /run/media/root/15fb790a-9d68-4d4c-a988-58f1da7322dc/T01P00_0.vob /run/media/root/15fb790a-9d68-4d4c-a988-58f1da7322dc/Slit-scan.mp4 /run/media/root/15fb790a-9d68-4d4c-a988-58f1da7322dc/shinjiro /tmp

real 0m12.665s
user 0m0.013s
sys 0m3.953s
sh-4.2#
This calculates to 127MB/s from the usb devices in raid0 to /tmp in the preospec archlinux raid0 system all in btrfs mode.

EDIT: 3//9/14... Upgraded all devices to linux-3.13.6-1-X86_64 and all is well. USB devices performing correctly and the boot time is ~2 secs.

Performed the following test on the transfer of 1.1GB file in USB 3.0 btrfs pair while mounted in prospec btrfs raid0 .....
sh-4.2# dd if=/run/media/root/15fb790a-9d68-4d4c-a988-58f1da7322dc/T01P00_0.vob of=/dev/null bs=1M count=1024
1024+0 records in
1024+0 records out
1073741824 bytes (1.1 GB) copied, 8.88375 s, 121 MB/s
sh-4.2#

This speed is that of the prospec pair from a pcie usb3.0 btrfs pair of compact flash ......devices limited to udma 5 speeds as applicable.//

No compression has been applied to any of the devices.

EDIT: 3/11/14..Installed sdhc 8GB mode 4 devices in SDHC slots on the two 3.0 pcie devices. Performed ...mkfs.btrfs -d raid0 ...on the devices.

Installed 2GB of data therein.

Rebooted and opened both the new devices and the adata raid devices in the pcie usb3.0 adapters and simultaneously displayed the contents of each pair.

The hdparm read speed is ~15MB/s with the sdhc devices whereas the adata devices read speed is 33MB/s.

The following data is obtained in fi show:

sh-4.2# btrfs fi show
Label: none uuid: fdf5f2d3-b078-41ab-a56c-88ea58e0e91b
Total devices 2 FS bytes used 5.25GiB
devid 1 size 14.92GiB used 4.03GiB path /dev/sda
devid 2 size 14.92GiB used 4.01GiB path /dev/sdd

Label: none uuid: 7012f04e-55fd-4475-8fbe-9536aff0ee0d
Total devices 2 FS bytes used 1.75GiB
devid 1 size 7.40GiB used 2.52GiB path /dev/sdp
devid 2 size 7.40GiB used 2.50GiB path /dev/sdk

Label: none uuid: 15fb790a-9d68-4d4c-a988-58f1da7322dc
Total devices 2 FS bytes used 1.75GiB
devid 1 size 14.92GiB used 2.03GiB path /dev/sdg
devid 2 size 14.92GiB used 2.01GiB path /dev/sdl

Label: none uuid: 036510c8-20c4-4a3e-9497-9ac01aab9588
Total devices 1 FS bytes used 2.46GiB
devid 1 size 29.82GiB used 5.04GiB path /dev/sde

Label: none uuid: d518c16c-59b5-44d8-a59c-4fab409e73e9
Total devices 1 FS bytes used 4.57GiB
devid 1 size 29.82GiB used 7.04GiB path /dev/sdb

Btrfs v3.12
sh-4.2#
This data represents all btrfs devices installed on my system.

I assume I can utilize another pair of sdhc ports on the 3.0 usb adapter for further raid0 pair performance.
I could also add devices to the raid0 to advance its capacity but make it modify to raid1 or raid10 depending on the number of devices added,

This permits a versatile application of compact flash devices easily and quickly programmed in btrfs mode in a mounted system.
The data installed to the sdhc devices is very quickly displayed in the /run/media/root mode by selection in the xfce4 menu of filesystems.
Wow!!!
.EDIT:3/13/14....Built four device btrfs raid10 using four 16GB devices...two devices are compact flash and two devices are 16GB mode 10 80MB/s rated.

These devices are mounted in the 3.0 adater connected to the pcie 3.0 interface adapter from a 100mhz bus on the mobo.

The following data was collected to establish the performance of the raid10 thusly produced(knowing that it is bandwidth limited)............

sh-4.2# btrfs fi show
Label: none uuid: fdf5f2d3-b078-41ab-a56c-88ea58e0e91b
Total devices 2 FS bytes used 7.32GiB
devid 1 size 14.92GiB used 5.03GiB path /dev/sda
devid 2 size 14.92GiB used 5.01GiB path /dev/sdd

Label: none uuid: 38d5bbc3-a81a-46e1-a2fd-83ab631508c6
Total devices 4 FS bytes used 1.75GiB
devid 1 size 14.84GiB used 2.02GiB path /dev/sdi
devid 2 size 14.84GiB used 2.00GiB path /dev/sdn
devid 3 size 14.92GiB used 1.01GiB path /dev/sdg
devid 4 size 14.92GiB used 1.01GiB path /dev/sdl

Label: none uuid: 7012f04e-55fd-4475-8fbe-9536aff0ee0d
Total devices 2 FS bytes used 1.75GiB
devid 1 size 7.40GiB used 1.78GiB path /dev/sdp
devid 2 size 7.40GiB used 1.78GiB path /dev/sdk

Label: none uuid: d518c16c-59b5-44d8-a59c-4fab409e73e9
Total devices 1 FS bytes used 4.57GiB
devid 1 size 29.82GiB used 7.04GiB path /dev/sdb

Label: none uuid: 036510c8-20c4-4a3e-9497-9ac01aab9588
Total devices 1 FS bytes used 2.46GiB
devid 1 size 29.82GiB used 5.04GiB path /dev/sde

Btrfs v3.12
sh-4.2# hdparm -t /dev/sdi /dev/sdn /dev/sdg /dev/sdl

/dev/sdi:
Timing buffered disk reads: 94 MB in 3.04 seconds = 30.90 MB/sec

/dev/sdn:
Timing buffered disk reads: 94 MB in 3.03 seconds = 31.03 MB/sec

/dev/sdg:
Timing buffered disk reads: 100 MB in 3.05 seconds = 32.78 MB/sec

/dev/sdl:
Timing buffered disk reads: 96 MB in 3.02 seconds = 31.82 MB/sec

sh-4.2# mount /dev/sdl /mnt/md
sh-4.2# cd /mnt/md

sh-4.2# btrfs fi balance start /mnt/md
Done, had to relocate 6 out of 6 chunks
sh-4.2# btrfs fi df /mnt/md

Data, RAID10: total=2.00GiB, used=1.75GiB
System, RAID1: total=32.00MiB, used=16.00KiB
Metadata, RAID1: total=1.00GiB, used=1.98MiB
sh-4.2#

The data indicates a four device raid10 storage capability has been established and that all four devices perform at 30MB/s in hdparm read mode.
When transferring the 2GB data into the raid10, most of the data transferred at 40MB/sec from the prospec mounted archlinux system by copy and paste.
There is data provided initially in this post which illustrates the full complement of devices presently connected to the system. These are displayed in the filemanager listing of "DEVICES". Any of the four devices in the raid10 array can be used to mount the raid in /mnt/md.

One test made with two 16GB sdhc devices involved using only one 3.0 adapter to house both devices. It performed in a normal fashion, indicating that raid0 operates differently from mdadm software raid AFAICT.

I am able to remove deices from the adapter while ionline and insert alternate devices having btrfs raid0 and recognize the new data thusly provided.

I have yet to utilize a normal 2.0 device in the mix to check that possible mode.

Fun!!!

EDIT: 3/14/14...Experimenting with mdadm in usb 3.0 adapter interface.

Established two compact flash devices of 533x 16GB in mdadm raid0 /dev/md127 and established two sdhc 16GB mode 10 devices in mdadm raid0 as /dev/md126..
Used mkfs.btrfs to establish mountable devices. Ran tests on each with the following results:

sh-4.2# btrfs fi show
Label: none uuid: fdf5f2d3-b078-41ab-a56c-88ea58e0e91b
Total devices 2 FS bytes used 7.62GiB
devid 1 size 14.92GiB used 5.03GiB path /dev/sda
devid 2 size 14.92GiB used 5.01GiB path /dev/sdb

Label: none uuid: 6444e612-8bd7-4a21-acc3-2d54c560a615
Total devices 1 FS bytes used 112.00KiB
devid 1 size 29.83GiB used 2.04GiB path /dev/md126

Label: none uuid: ab7f5d77-09c2-4b29-aef1-a898e48a5341
Total devices 1 FS bytes used 112.00KiB
devid 1 size 29.66GiB used 2.04GiB path /dev/md127

Btrfs v3.12
sh-4.2# hdpar -t /dev/md126 /dev/md127
sh: hdpar: command not found
sh-4.2# hdparm -t /dev/md126 /dev/md127

/dev/md126:
Timing buffered disk reads: 258 MB in 3.01 seconds = 85.60 MB/sec

/dev/md127:
Timing buffered disk reads: 248 MB in 3.00 seconds = 82.63 MB/sec
sh-4.2#
sh-4.2# btrfs fi df /mnt/md
Data, single: total=8.00MiB, used=256.00KiB
System, DUP: total=8.00MiB, used=16.00KiB
System, single: total=4.00MiB, used=0.00
Metadata, DUP: total=1.00GiB, used=112.00KiB
Metadata, single: total=8.00MiB, used=0.00

sh-4.2# mount /dev/md127 /mnt/md1
sh-4.2# cd /mnt/md1
sh-4.2# btrfs fi df /mnt/md1
Data, single: total=8.00MiB, used=256.00KiB
System, DUP: total=8.00MiB, used=16.00KiB
System, single: total=4.00MiB, used=0.00
Metadata, DUP: total=1.00GiB, used=112.00KiB
Metadata, single: total=8.00MiB, used=0.00
sh-4.2#

At present there is no data in either raid array. Each array uses a port on either of the two adapters bothe adapters connected to a two-port usb 3.0 adapter which obtains input from a 100mhz bus.

Both arrays appear in cat /proc/mdstat and in btrfs fi show.

A third pair of sdhc devices could be mdadm raided with the present setup. This effort appears below:

sh-4.2# btrfs fi show
Label: none uuid: fdf5f2d3-b078-41ab-a56c-88ea58e0e91b
Total devices 2 FS bytes used 7.62GiB
devid 1 size 14.92GiB used 5.03GiB path /dev/sda
devid 2 size 14.92GiB used 5.01GiB path /dev/sdb

Label: none uuid: 6444e612-8bd7-4a21-acc3-2d54c560a615
Total devices 1 FS bytes used 384.00KiB
devid 1 size 29.83GiB used 2.04GiB path /dev/md126

Label: none uuid: ab7f5d77-09c2-4b29-aef1-a898e48a5341
Total devices 1 FS bytes used 384.00KiB
devid 1 size 29.66GiB used 2.04GiB path /dev/md127

Label: none uuid: 7012f04e-55fd-4475-8fbe-9536aff0ee0d
Total devices 2 FS bytes used 1.75GiB
devid 1 size 7.40GiB used 1.78GiB path /dev/sdg
devid 2 size 7.40GiB used 1.78GiB path /dev/sdl

Btrfs v3.12
sh-4.2# mdadm --create /dev/md125 --level=0 --raid-devices=2 /dev/sdg /dev/sdl
mdadm: Defaulting to version 1.2 metadata
mdadm: array /dev/md125 started.
sh-4.2# mkfs.btrfs /dev/md125

WARNING! - Btrfs v3.12 IS EXPERIMENTAL
WARNING! - see http://btrfs.wiki.kernel.org before using

Turning ON incompat feature 'extref': increased hardlink limit per file to 65536
fs created label (null) on /dev/md125
nodesize 16384 leafsize 16384 sectorsize 4096 size 14.80GiB
Btrfs v3.12
sh-4.2# mount /dev/md125 /mnt/md
sh-4.2# cd /mnt/md
sh-4.2# btrfs fi df /mnt/md
Data, single: total=8.00MiB, used=256.00KiB
System, DUP: total=8.00MiB, used=16.00KiB
System, single: total=4.00MiB, used=0.00
Metadata, DUP: total=1.00GiB, used=112.00KiB
Metadata, single: total=8.00MiB, used=0.00
sh-4.2# hdparm -t /dev/md125

/dev/md125:
Timing buffered disk reads: 106 MB in 3.01 seconds = 35.27 MB/sec
sh-4.2# These are mode 4 devices of 8 GB sdhc.

Thus, three btrfs mdadm raid0 arrays using full-capacity devices are established in my system on USB 3.0 adapters. They appear in both cat/proc/mdstat and btrfs fi show.

Further data shows good transfer rate of 1.6GB to /tmp file.....
sh-4.2# time cp /run/media/root/ab7f5d77-09c2-4b29-aef1-a898e48a5341/T01P00_0.vob /run/media/root/ab7f5d77-09c2-4b29-aef1-a898e48a5341/Slit-scan.mp4 /run/media/root/ab7f5d77-09c2-4b29-aef1-a898e48a5341/shinjiro /tmp

real 0m18.785s
user 0m0.027s
sys 0m3.717s
sh-4.2#

This was timed and the result is 85 MB/s transfer rate. This compares with the write results obtained during install to mdadm of 2GB of data(including the data written to /tmp)which showed a steady write speed of 40MB/sec indicating that speed is provided in thearrangement.

Further experimental wildness is deicted below:

sh-4.2# cat /proc/mdstat
Personalities : [raid0]
md125 : active raid0 sdm[1] sdh[0]
15515648 blocks super 1.2 512k chunks

md126 : active raid0 sdd[0] sdi[1]
31276032 blocks super 1.2 512k chunks

md127 : active raid0 sdf[0] sdk[1]
31099904 blocks super 1.2 512k chunks

unused devices: <none>
sh-4.2# mount /dev/md126 /mnt/md
sh-4.2# btrfs fi balance start /mnt/md
Done, had to relocate 7 out of 7 chunks
sh-4.2# umount /mnt/md
sh-4.2# mount /dev/md125 /mnt/md
sh-4.2# btrfs fi balance start /mnt/md
Done, had to relocate 7 out of 7 chunks
sh-4.2# umount /mnt/md
sh-4.2# mount /dev/md127 /mnt/md
sh-4.2# btrfs fi balance start /mnt/md
Done, had to relocate 7 out of 7 chunks
sh-4.2#
I might have to change the chunk size for a look-see what happens then!!! The balance procedure requires two devices to balance so it splits the mdadm while in /mnt/md format. Another wild experiment!!!!

So there are 3 mdadm pairs in the two adapters connected to a pcie 1x 2 port usb 3.0 interface to a 100mhz bus. They act independently but don't yet know if they can interplay with each other. It certainly doesn&0039;t seem reasonable.......but this is btrfs and it is fabulous!!!!

Well, it sure is interesting that I can interplay with the mdadm raids with some success. Timed transfer from /mounted raids of 320MB file was made at 84MB/s but a file of size 1.1 GB transferred at 26MB/s. I have to dream up some more things to play with now!!!!

EDIT 3/15/14 ...Changed mdadm chunk size in /dev/md126 to 4096. Performed mkfs.btrfs and mounted the result in /mnt/md. Ran balance procedure as before. Performed times cp of 1600MB files to /tmp. Results below indicate 88MB/s using the prospec devices in archlinux OS boot. The data appears below:

sh-4.2# cat /proc/mdstat
Personalities : [raid0]
md126 : active raid0 sdd[1] sdi[0]
31268864 blocks super 1.2 4096k chunks

md127 : active raid0 sdk[1] sdf[0]
31099904 blocks super 1.2 512k chunks

md125 : active raid0 sdh[0] sdm[1]
15515648 blocks super 1.2 512k chunks

unused devices: <none>
sh-4.2# mount /dev/md126 /mnt/md
sh-4.2# cd /mnt/md
sh-4.2# time cp /mnt/md/T01P00_0.vob /mnt/md/Slit-scan.mp4 /mnt/md/shinjiro /tmp

real 0m18.153s
user 0m0.043s
sys 0m3.683s
sh-4.2#

This does not indicate any significant improvement in transfer speed, basically the storage device mounted in /mnt/md ihas data transferred via the OS which has not been modified.

EDIT: 3/16/14...Performed ....mkfs.btrfs -f /dev/md127 /dev/md126 to create a combined four device arrangement. This produced a mountable btrfs device in /mnt/md which was empty of any data. This indicated that the btrfs system was established....copied 2GB of files into /mnt/md. Exited and ran ...btrfs fi show... with the following results:

sh-4.2# mount /dev/md126 /mnt/md
sh-4.2# btrfs fi show
Label: none uuid: fdf5f2d3-b078-41ab-a56c-88ea58e0e91b
Total devices 2 FS bytes used 7.04GiB
devid 1 size 14.92GiB used 6.03GiB path /dev/sda
devid 2 size 14.92GiB used 6.01GiB path /dev/sdc

Label: none uuid: f0293e86-20dc-4cc5-8f04-29001a686223
Total devices 2 FS bytes used 1.75GiB
devid 1 size 29.82GiB used 2.03GiB path /dev/md126
devid 2 size 29.66GiB used 2.01GiB path /dev/md127

Label: none uuid: 836d0167-8205-4633-8b58-601db4f888d7
Total devices 1 FS bytes used 1.77GiB
devid 1 size 14.80GiB used 3.31GiB path /dev/md125

Btrfs v3.12
sh-4.2#

Thus the raid0 pairs in the mdadm raids are now combined in a four device btrfs system. The capacity is full size for the four devices thus it approaches 64MB.

What else can I do to mess with btrfs file system? Cat /proc/mdstat still identifies the mdadm raid0 arrays as /md127 and /md126.
Either of the two mdadm devices can be used to mount the btrfs data installed into the paired mdadm devices.

The chunk sizes provided in the creation of the mdadm raid0 arrays is different in each device. The btrfs system was created on top of that disparity.

Mounting the paired mdadm devices shows 59GB of free space following the copy of 2GB data. This suggests two levels of btrfs of 2GB each exist in the filesystem of total 64GB. Furthermore, 32GB of the total is sdhc devices and 32Gb is compact flash devices..

What next?

Added EDITS:3/15/14...Performed the following procedures on the new paired raid0 devices with the results shown:

sh-4.2# systemd-analyze
Startup finished in 1.212s (kernel) + 954ms (userspace) = 2.166s
sh-4.2# mount /dev/md127 /mnt/md
sh-4.2# cd /mnt/md
sh-4.2# btrfs fi balance start /mnt/md
Done, had to relocate 6 out of 6 chunks
sh-4.2# btrfs fi df /mnt/md
Data, RAID0: total=2.00GiB, used=1.75GiB
System, RAID1: total=32.00MiB, used=16.00KiB
Metadata, RAID1: total=1.00GiB, used=2.00MiB
sh-4.2#

This arrangement provides ~64GB of storage using full-capacity of the devices. I assume I can utilize any compatible devices within this framework by replacing the existing devices with other sdhc or compact flash devices to then utilize further 64GB storage(removable). It may require balancing to do so.

As shown the systemd-analyze boot data is still ~2 seconds. Thus, utilizing USB pcie devices for storage does not affect boot time.

EDIT: St. Paddy's day 3/17/14... Ran hdparm on the raid pairs consisting of two 16GB adata 533x compact flash and two 16GB sdhc mode 10 devices. The results are supplied below;
sh-4.2# hdparm -tT /dev/md126 /dev/md127

/dev/md126:
Timing cached reads: 2308 MB in 2.00 seconds = 1154.39 MB/sec
Timing buffered disk reads: 254 MB in 3.01 seconds = 84.52 MB/sec

/dev/md127:
Timing cached reads: 2350 MB in 2.00 seconds = 1175.11 MB/sec
Timing buffered disk reads: 250 MB in 3.02 seconds = 82.79 MB/sec
sh-4.2#

These figures are for usb 3.0 connected raid pairs with btrfs filesystem. The filesystem is selected by the /md126 pair for mounting and it is listed as a 31gb volume in the menu DEVICES.

Transfer speed tests for data in the 64GB mount of the two raid pairs reveals only 40MB/s. The raid pairs provided by the set-up are data only and are not bootable OS pairs nor could they be booted from pcie in any case.

Either device, md126 or md127, can be used to mount the 64GB pair from command line, but selection

EDIT; 10/3/14 IT SEEMS THAT MY EDITS HAVE BEEN REMOVED FOR BTRFS ACTIVITY AT N6RE.

i AM PRESENTLY IN LINUX-3.16.3-1 AND HAVE UPGRADED ALL BTRFS SYSTEMS SUCCESSFULLY TO THAT LEVEL. tHE BASH REPAIR ACTIVITY HAS BEEN INCLUDED AND AM WONDERING IF THAT IS THE CAUSE OF MY DATA LOSS IN THIS POST. tHERE HAS BEEN AN INCREASE OF ACTIVITY FOR THIS POST RECENTLY WHICH WOULD ACCOUNT FOR MALICIOUS ACTIVITY.

EDIT: 10/7/14...Upgraded to linux-3.16.4-1-x86....with no problems. I have received a bashing with several upgrades to bash as well. The present btrfs arrangement is shown below:
sh-4.3# btrfs fi show
Label: none uuid: c3061fa3-5476-42bf-a8f0-6a0c11e2c8e1
Total devices 1 FS bytes used 3.75GiB
devid 1 size 14.92GiB used 8.04GiB path /dev/sda

Label: none uuid: a3659970-8231-4cf8-9f5e-ed2aab05ea51
Total devices 1 FS bytes used 1.85GiB
devid 1 size 14.92GiB used 4.04GiB path /dev/sdc

Label: none uuid: 4aabd786-84fb-4672-bd90-05c2e962a004
Total devices 1 FS bytes used 6.97GiB
devid 1 size 29.82GiB used 29.82GiB path /dev/sdd

Label: none uuid: aa2b8760-a417-4630-aedb-9ee6e0ea867b
Total devices 1 FS bytes used 7.66GiB
devid 1 size 14.92GiB used 14.92GiB path /dev/sdb

Label: none uuid: a0004298-149e-4bce-ae79-ada1d3bfef8d
Total devices 1 FS bytes used 2.54GiB
devid 1 size 29.82GiB used 21.04GiB path /dev/sde

Btrfs v3.16.2
sh-4.3#
Thus, my systems are all single devices. The boot times are affected by systemd such that some boots are as long as 15 sec but reboots occur at ~5 sec. One system is on USB sdhc 16GB and it boots in less than ten seconds.All systems perform well in archlinux using otter-browser, firefox and jumanji. Some systems use thunderbird ...some use roudcube,,,and others use squirrelmail. Kdenlive and/or openshot for video editing and the very nice mpv for videos. All systems are root only and x86 type...mostly compact flash devices with sata adapters. I have one pair of maxell 16GB udma 400x provided as a mdadm raid0 array which reads at 172MB/s . It resides in a pair of sata adapters when it displaces the systems therein. Gvfs provides easy access to that array as well as the rest of the systems as installed in their sata ports. One system is connected to the only IDE port provided in my gcm945 mobo. The cpu fan is the only noise producing device in the computer system.

EDIT 1014/14...Installed aura program today and enjoy it very much. It upgraded my system's aur pkgs with one command and installed upgrades in one operation. Very fine setup indeed! Will have to study the man page for operating with this new toy!

EDIT: 10/21/14...Aura helped nicely to replace my kdenlive which failed to utilize audio in aditing some videos. The error report said sdl connection fail or some such. This version of kdenlive was downloaded from pacman and utilized mlt from community. Decided to try kdenlive and mlt from aur-git sources. Most of the depends were through pacman -S but some required additional aur packages. Aura -A (package name) performed very nicely with these packages, even to the point of downloading the needed depends. After finishing the builds, the kdenlive-git performed excellent joining of video and audio stereo files. Thus the SDL problem is inherent in the pacman -S kdenlive download. One area which is quirky is the naming of packages. One package is named qimageblitz....but i read it as gimageblitz with my excellent eyesight!!! Needless to say i tumbled to the error on reflection after querying for the package. So that's how it crumbles at times!!!!! Very pleased to get the kdenlive operating correctly....

EDIT: 10/23/14 --Upgraded to linux-3.17.1-1-x86...on six flash-based btrfs systems. Noted the intel-ucode upgrade and installed same in one system. I assume this applies to an upcoming upgrade rather than this one. Also installed openshot which responds well to all video files with audio present in all cases as differed from kdenlive.

EDIT2: Noted the post by Trilby re 4 second boot time. Noted that my 4 second boot time began in earlier kernel upgrades as opposed to Trilby having it disappear with previous kernel. Thus his activities in the previous kernel boot versus the present 3.17 might give a clue to the cause since it is always there for me in many upgrades from 3.15 on. This suggests iit might be systemd related in the kernel setup since systemd has been upgraded many times in recent months. Why his system can fall back to previous kernel to correct the excessive time is a puzzle since my systems have had the problem for many upgrades. All of my systems (6) have the 4 sec delay.

EDIT: 10/29/14....Noted a change in systemd-analyze which raised my boot time from 5+ secs to 24 secs mostly in user space. This seems to be the addition of new systemd parameters introduced by systemd growth. Suspect this is the result of increasing diversity in archlinux caused by multi-core mobos and multi-boot device formats.

The original install of my root only btrfs system booted in kernel in ~1.083 seconds and user space in ~1 second for a total of 2+ seconds and this was real time.

I cannot expect to recover that boot time in view of the growth pattern in systemd.

Perhaps I will have to reject many of the new systemd controls newly initiated to regain short boot time.

I cannot reduce kernel boot time in spite of many tries to control tsc/hpet/pit so it remains at ~ 4 secs. Tsc seems to cause 3 seconds of delay when switched from HPET in the dmesg sequence.

So my idea of KISS using fast boot time and root only is fast disappearing. And the system performance has deteriorated as well with hangs showing in activities including this

EDIT: 11/2/14....Upgrade to linux-3.17.2-1 successful on all six systems. Utilized intel-ucode and /boot/intrl-ucode.img params as well as mkinitcpio-lz4 compression on all systems. All are booting with near 5 sec boot time. However, boot message says ....initrd is corrupted.... but am not aware as to why that message appears during boot time with quiet enabled. Systemd-analyze shows approx 6 secs in USB system with sdhc 16GB mode10 device in usrobotics 3.0 adapter.

EDIT: 11/10/14 ...Maxell 400x udma compact Flash devices fail to operate correctly in archlinux. These units are 5 years old and have been in use for most of the time in root only mode. Thus two devices are no longer viable for booting archlinux. Newer devices continue to operate in root only mode with no problem. Boot time in IDE mode is ~5 secs. Most devices are 16GB in size, two are 32GB. The boot time for this operating device 16GB is shown below;
sh-4.3# systemd-analyze
Startup finished in 413ms (kernel) + 3.423s (initrd) + 1.351s (userspace) = 5.187s
sh-4.3#

Data for the units now in service is shown below;

sh-4.3# btrfs fi show
Label: none uuid: 7dddf155-dcac-4ee6-ab4f-a97ad822bdb8
Total devices 1 FS bytes used 4.30GiB
devid 1 size 14.92GiB used 8.04GiB path /dev/sda

Label: none uuid: a3659970-8231-4cf8-9f5e-ed2aab05ea51
Total devices 1 FS bytes used 3.61GiB
devid 1 size 14.92GiB used 7.04GiB path /dev/sdb

Btrfs v3.17.1
sh-4.3#

EDIT: 11/13/14....Installed an ocx-vertex2 60GB SSD device into my 945GCM 2C mobo. Installed archlinux to full-capacity btrfs with no problem. Editing this post with that system operating. I submit several test results regarding this device performance and details from btrfs.
sh-4.3# btrfs fi show
Label: none uuid: f76d984a-dc5e-4641-bf58-aa886013312e
Total devices 1 FS bytes used 4.05GiB
devid 1 size 55.90GiB used 7.04GiB path /dev/sdb

Label: none uuid: 7dddf155-dcac-4ee6-ab4f-a97ad822bdb8
Total devices 1 FS bytes used 3.50GiB
devid 1 size 14.92GiB used 9.04GiB path /dev/sda

Label: none uuid: ec031c1f-de69-44f3-b8bc-5f58ddc66df3
Total devices 2 FS bytes used 8.87GiB
devid 1 size 29.83GiB used 6.03GiB path /dev/md126
devid 2 size 29.66GiB used 6.01GiB path /dev/md127

Btrfs v3.17.1

Read speed tests follow;

sh-4.3# hdparm -t /dev/sda /dev/sdb /dev/md126 /dev/md127

/dev/sda:
Timing buffered disk reads: 250 MB in 3.00 seconds = 83.23 MB/sec

/dev/sdb:
Timing buffered disk reads: 590 MB in 3.01 seconds = 196.11 MB/sec

/dev/md126:
Timing buffered disk reads: 244 MB in 3.02 seconds = 80.78 MB/sec

/dev/md127:
Timing buffered disk reads: 258 MB in 3.01 seconds = 85.79 MB/sec
sh-4.3#

Three btrfs systems are tested, sda is a 16GB prospec cf card mounted in an IDE adapter pata port. Device /sdb is the ocx-vertex2 device of 60GB. The md devices are connected via USRobotics usb3.0 adapter for two raid arrays ....one is a pair of sdhc 16GB mode 10 devices and the other is a pair of 16GB 533x cf cards. These connect thru the robotics adapter to a pcie card providing 3.0 usb ports (2) from a 100mhz bus in the mobo. These md devices are raid0 and joined with btrfs as a two device storage arrangement of ~60GB. This storage arrangement is installed in tmpfs by my system.

A transfer test of a video file of 1.1GB gave a transfer time of 11.381 seconds which computes to 96.65 MB/s for the ocx-vertex device. I assume this to be the write speed of the device for that file which was located in /var and written to Desktop.

Of some interest is the read speed results if made while booted in the /dev/sda. These results indicate that the device utilized to perform the test are affected by that device's parameters. For example, the ocx-vertex reading when tested while booted in /dev/sda was 136MB/s.

It would seem that the device being used to measure read speed has to be faster than the read speed of the tested device.

The speeds shown for the md devices is quite surprising in view of the USB connection altho USB3.0 should permit higher speed results. In this case, however, the bus limits the results due to the 100mhz bus connection.

The ocx-vertex2 device derives its bus from SATAII which in this core2 duo is near 1100MB/s while using x86_64 operating system with interleaved cores. I suspect if using i686 the source would be doubled IMHOP.

Now need to try a pair of full-capacity ocx-vertex devices in btrfs raid0.

EDIT: The following systemd-analyze shows some improvement in boot time....
sh-4.3# systemd-analyze
Startup finished in 333ms (kernel) + 3.294s (initrd) + 1.037s (userspace) = 4.665s
sh-4.3#

I would entertain a method to reduce the initrd boot time somehow!!!! n6re@mminternet.com...

EDIT 11/16/14...Performed a btrfs raid0 install from a booted archlinux OS as described below;

The /dev/sda is maxell 400x 16GB udma device with arch installed. /dev /sdb is another maxell 16gb device 400x udma to be added to the btrfs raid array as shown below;

#mkfs.btrfs -f /dev/sdb
#mount /dev/sda /mnt/md
#btrfs device add /dev/sdb /mnt/md
#cd /mnt/md
#btrfs balance start -f -dconvert=raid0 /mnt/md

After some time the raid0 pair were joined. The data shows it present below;

sh-4.3# btrfs fi show
Label: none uuid: 55152396-4852-4d45-9036-40f02340742f
Total devices 2 FS bytes used 2.43GiB
devid 1 size 14.89GiB used 6.03GiB path /dev/sda
devid 2 size 14.89GiB used 4.00GiB path /dev/sdb

Label: none uuid: 74dd4312-a47c-40a2-82c2-bc195743f720
Total devices 1 FS bytes used 8.08GiB
devid 1 size 14.84GiB used 11.04GiB path /dev/sdh

Label: none uuid: ec031c1f-de69-44f3-b8bc-5f58ddc66df3
Total devices 2 FS bytes used 8.87GiB
devid 1 size 29.83GiB used 6.03GiB path /dev/md127
devid 2 size 29.66GiB used 6.01GiB path /dev/md126

Label: none uuid: b395d286-85f4-4059-8621-7187c6d5d9d6
Total devices 1 FS bytes used 354.98MiB
devid 1 size 14.84GiB used 3.04GiB path /dev/sdm

Btrfs v3.17.1
sh-4.3#

EDIT: 11/17/14...Upgraded to linux-3.17.2-1-x86...with no problems. Also converted a silicon power 32GB pair of 32GB devices to raid0 btrfs in the same manner as before. Hoping to do the same with the ocz device soon altho I do not expect any great performance improvement. The devices from ocz are used devices. They are utilized in full-capacity mode and as with all my systems root only. Everything is experimental. Have yet to get the second ocz unit, perhaps tmw.

EDIT2...Systemd-analyze results for ocz device;
sh-4.3# systemd-analyze
Startup finished in 299ms (kernel) + 3.255s (initrd) + 1.248s (userspace) = 4.804s
sh-4.3#

I would entertain ways to reduce initrd boot time.

EDIT 11/19/14...Installed a second ocz vertex2 device and added it to the existing btrfs OS with the device add procedure and balance -dconvert=raid0.

The procedure generated a btrfs pair in raid0 which boots in systemd as follows;

sh-4.3# systemd-analyze
Startup finished in 283ms (kernel) + 3.246s (initrd) + 1.204s (userspace) = 4.734s
sh-4.3#

sh-4.3# btrfs fi show
Label: none uuid: f76d984a-dc5e-4641-bf58-aa886013312e
Total devices 2 FS bytes used 4.31GiB
devid 1 size 55.90GiB used 16.03GiB path /dev/sdb
devid 2 size 55.90GiB used 14.00GiB path /dev/sda

Btrfs v3.17.1

The system boots on either /dev/sda .../dev/sdb.

A 600MB+ video file was transferred from /var to Desktop @ 153MB/sec.

Hdparm reports these devices as udma6. This is refleted in dmesg as well.

EDIT: 11/20/14....The following data represents the ssd parameters for the used ocz vertex2 devices:
sh-4.3# hdparm -i /dev/sda /dev/sdb

/dev/sda:

Model=OCZ-VERTEX2, FwRev=1.32, SerialNo=OCZ-UXH37EO7Z536H902
Config={ Fixed }
RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=4
BuffType=unknown, BuffSize=unknown, MaxMultSect=16, MultSect=16
CurCHS=16383/16/63, CurSects=16514064, LBA=yes, LBAsects=117229295
IORDY=on/off, tPIO={min:120,w/IORDY:120}, tDMA={min:120,rec:120}
PIO modes: pio0 pio1 pio2 pio3 pio4
DMA modes: mdma0 mdma1 mdma2
UDMA modes: udma0 udma1 udma2 udma3 udma4 udma5 *udma6
AdvancedPM=no WriteCache=enabled
Drive conforms to: unknown: ATA/ATAPI-2,3,4,5,6,7

* signifies the current active mode

/dev/sdb:

Model=OCZ-VERTEX2, FwRev=1.32, SerialNo=OCZ-616543F884M6H305
Config={ Fixed }
RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=4
BuffType=unknown, BuffSize=unknown, MaxMultSect=16, MultSect=16
CurCHS=16383/16/63, CurSects=16514064, LBA=yes, LBAsects=117229295
IORDY=on/off, tPIO={min:120,w/IORDY:120}, tDMA={min:120,rec:120}
PIO modes: pio0 pio1 pio2 pio3 pio4
DMA modes: mdma0 mdma1 mdma2
UDMA modes: udma0 udma1 udma2 udma3 udma4 udma5 *udma6
AdvancedPM=no WriteCache=enabled
Drive conforms to: unknown: ATA/ATAPI-2,3,4,5,6,7

* signifies the current active mode

sh-4.3# hdparm -I /dev/sda /dev/sdb

/dev/sda:

ATA device, with non-removable media
Model Number: OCZ-VERTEX2
Serial Number: OCZ-UXH37EO7Z536H902
Firmware Revision: 1.32
Transport: Serial
Standards:
Used: unknown (minor revision code 0x0028)
Supported: 8 7 6 5
Likely used: 8
Configuration:
Logical max current
cylinders 16383 16383
heads 16 16
sectors/track 63 63
--
CHS current addressable sectors: 16514064
LBA user addressable sectors: 117229295
LBA48 user addressable sectors: 117229295
Logical Sector size: 512 bytes
Physical Sector size: 512 bytes
Logical Sector-0 offset: 0 bytes
device size with M = 1024*1024: 57240 MBytes
device size with M = 1000*1000: 60021 MBytes (60 GB)
cache/buffer size = unknown
Nominal Media Rotation Rate: Solid State Device
Capabilities:
LBA, IORDY(can be disabled)
Queue depth: 32
Standby timer values: spec'd by Standard, no device specific minimum
R/W multiple sector transfer: Max = 16 Current = 16
DMA: mdma0 mdma1 mdma2 udma0 udma1 udma2 udma3 udma4 udma5 *udma6
Cycle time: min=120ns recommended=120ns
PIO: pio0 pio1 pio2 pio3 pio4
Cycle time: no flow control=120ns IORDY flow control=120ns
Commands/features:
Enabled Supported:
* SMART feature set
Security Mode feature set
* Power Management feature set
* Write cache
* Look-ahead
* Host Protected Area feature set
* WRITE_BUFFER command
* READ_BUFFER command
* NOP cmd
* DOWNLOAD_MICROCODE
SET_MAX security extension
* 48-bit Address feature set
* Mandatory FLUSH_CACHE
* FLUSH_CACHE_EXT
* SMART error logging
* SMART self-test
* General Purpose Logging feature set
* WRITE_{DMA|MULTIPLE}_FUA_EXT
* 64-bit World wide name
* IDLE_IMMEDIATE with UNLOAD
* WRITE_UNCORRECTABLE_EXT command
* Segmented DOWNLOAD_MICROCODE
* Gen1 signaling speed (1.5Gb/s)
* Gen2 signaling speed (3.0Gb/s)
* Native Command Queueing (NCQ)
* Host-initiated interface power management
* Phy event counters
DMA Setup Auto-Activate optimization
Device-initiated interface power management
* Software settings preservation
* SMART Command Transport (SCT) feature set
* SCT Write Same (AC2)
* SCT Error Recovery Control (AC3)
* SCT Features Control (AC4)
* SCT Data Tables (AC5)
* Data Set Management TRIM supported (limit 1 block)
* Deterministic read data after TRIM
Security:
supported
not enabled
not locked
not frozen
not expired: security count
supported: enhanced erase
400min for SECURITY ERASE UNIT. 400min for ENHANCED SECURITY ERASE UNIT.
Logical Unit WWN Device Identifier: 5e83a97fd7ff20e4
NAA : 5
IEEE OUI : e83a97
Unique ID : fd7ff20e4
Checksum: correct

/dev/sdb:

ATA device, with non-removable media
Model Number: OCZ-VERTEX2
Serial Number: OCZ-616543F884M6H305
Firmware Revision: 1.32
Transport: Serial
Standards:
Used: unknown (minor revision code 0x0028)
Supported: 8 7 6 5
Likely used: 8
Configuration:
Logical max current
cylinders 16383 16383
heads 16 16
sectors/track 63 63
--
CHS current addressable sectors: 16514064
LBA user addressable sectors: 117229295
LBA48 user addressable sectors: 117229295
Logical Sector size: 512 bytes
Physical Sector size: 512 bytes
Logical Sector-0 offset: 0 bytes
device size with M = 1024*1024: 57240 MBytes
device size with M = 1000*1000: 60021 MBytes (60 GB)
cache/buffer size = unknown
Nominal Media Rotation Rate: Solid State Device
Capabilities:
LBA, IORDY(can be disabled)
Queue depth: 32
Standby timer values: spec'd by Standard, no device specific minimum
R/W multiple sector transfer: Max = 16 Current = 16
DMA: mdma0 mdma1 mdma2 udma0 udma1 udma2 udma3 udma4 udma5 *udma6
Cycle time: min=120ns recommended=120ns
PIO: pio0 pio1 pio2 pio3 pio4
Cycle time: no flow control=120ns IORDY flow control=120ns
Commands/features:
Enabled Supported:
* SMART feature set
Security Mode feature set
* Power Management feature set
* Write cache
* Look-ahead
* Host Protected Area feature set
* WRITE_BUFFER command
* READ_BUFFER command
* NOP cmd
* DOWNLOAD_MICROCODE
SET_MAX security extension
* 48-bit Address feature set
* Mandatory FLUSH_CACHE
* FLUSH_CACHE_EXT
* SMART error logging
* SMART self-test
* General Purpose Logging feature set
* WRITE_{DMA|MULTIPLE}_FUA_EXT
* 64-bit World wide name
* IDLE_IMMEDIATE with UNLOAD
* WRITE_UNCORRECTABLE_EXT command
* Segmented DOWNLOAD_MICROCODE
* Gen1 signaling speed (1.5Gb/s)
* Gen2 signaling speed (3.0Gb/s)
* Native Command Queueing (NCQ)
* Host-initiated interface power management
* Phy event counters
DMA Setup Auto-Activate optimization
Device-initiated interface power management
* Software settings preservation
* SMART Command Transport (SCT) feature set
* SCT Write Same (AC2)
* SCT Error Recovery Control (AC3)
* SCT Features Control (AC4)
* SCT Data Tables (AC5)
* Data Set Management TRIM supported (limit 1 block)
* Deterministic read data after TRIM
Security:
supported
not enabled
not locked
not frozen
not expired: security count
supported: enhanced erase
400min for SECURITY ERASE UNIT. 400min for ENHANCED SECURITY ERASE UNIT.
Logical Unit WWN Device Identifier: 5e83a97fb721c973
NAA : 5
IEEE OUI : e83a97
Unique ID : fb721c973
Checksum: correct
sh-4.3#

Performing the read test reveals a large difference in read speed;
sh-4.3# hdparm -Tt /dev/sda /dev/sdb

/dev/sda:
Timing cached reads: 2326 MB in 2.00 seconds = 1163.10 MB/sec
Timing buffered disk reads: 476 MB in 3.01 seconds = 158.30 MB/sec

/dev/sdb:
Timing cached reads: 2368 MB in 2.00 seconds = 1183.58 MB/sec
Timing buffered disk reads: 572 MB in 3.00 seconds = 190.59 MB/sec
sh-4.3#

The slower device is the added device. No idea why it is slower.

EDIT: 11/21/14 ...Installed another oczvertex2 device and installed archlinux. The resulting OS performs as noted below;

sh-4.3# systemd-analyze
Startup finished in 415ms (kernel) + 3.254s (initrd) + 954ms (userspace) = 4.624s
sh-4.3# btrfs fi show
Label: none uuid: 82e66cd7-09a9-467b-a5f5-d94abf216d10
Total devices 1 FS bytes used 1.86GiB
devid 1 size 55.90GiB used 4.04GiB path /dev/sdb

Label: none uuid: 7dddf155-dcac-4ee6-ab4f-a97ad822bdb8
Total devices 1 FS bytes used 3.62GiB
devid 1 size 14.92GiB used 9.04GiB path /dev/sda

Btrfs v3.17.1
sh-4.3# hdparm -Tt /dev/sdb

/dev/sdb:
Timing cached reads: 2356 MB in 2.00 seconds = 1178.22 MB/sec
Timing buffered disk reads: 622 MB in 3.00 seconds = 207.21 MB/sec
sh-4.3#

These are good indicators for an SSD in x86_64 mode with ~4GB installed.

It would seem advntageous to utilize this /sdb device with the other device running in raid0 with a read speed of 190MB/s.

EDIT: Tranfer test with 650MB file gave 276MB/s rate in btrfs raid0 pair of ocz vertex2 devices.
Hdparm cannot measure the read speed of the pair, only individual drives in the btrfs mode. This is in contrast to mdadm raid arrays. Using btrfs allows grub boot full-capacity devices without partitioning, boot to Desktop direct with no log-in and root only operation. Typical boot time is ~5 seconds or less.

SataII ports become sata only in my mobo if using a sata based drive (such as ocz vertex2) Thus, with my mobo, I cannot run compact flash devices from sata terminals if a sata drive is connected to any of the four sata ports. I can, however, use the pata IDE port of the mobo with an adapter use compact flash devices to boot or otherwise. Also, can utilize USB for booting or storage of compact flash. In the absence of sata drives, the sata ports become IDE based if used with IDE adapters. Mobo is gigabyte 945GCM 2C.

EDIT: 11/26/14 ....Ugraded the ocz vertex drives to linux-3.17.4-1-x86...with no problems. oot time for single ocz device is showb below;
sh-4.3# systemd-analyze
Startup finished in 419ms (kernel) + 3.232s (initrd) + 1.133s (userspace) = 4.785s
sh-4.3#

Expect to get another ocz drive soon.

EDIT: 12/3/14...With the fourth vertex2 installed, I performed device add with the single btrfs device.
After verifying the setup in fi show, I attempted to join the four devices with the commands....

Mount /dev/sdb /mnt/md
btrfs device add -f /dev/sde /mnt/md
Later on I performed a balance command while booted into the four device btrfs system:

btrfs balance /

The result is shown below:

sh-4.3# btrfs balance start /
Done, had to relocate 7 out of 7 chunks
sh-4.3# btrfs fi show
Label: none uuid: 073f4158-1b28-4a48-a1e9-9b7840102ea8
Total devices 4 FS bytes used 5.47GiB
devid 1 size 55.90GiB used 3.03GiB path /dev/sdb
devid 2 size 55.90GiB used 3.03GiB path /dev/sdc
devid 3 size 55.90GiB used 4.00GiB path /dev/sde
devid 4 size 55.90GiB used 4.00GiB path /dev/sdd

Label: none uuid: a3659970-8231-4cf8-9f5e-ed2aab05ea51
Total devices 1 FS bytes used 4.23GiB
devid 1 size 14.92GiB used 7.04GiB path /dev/sda

Label: none uuid: 7ca21c5b-e887-435a-9db3-1ad13b145aea
Total devices 2 FS bytes used 12.85GiB
devid 1 size 29.83GiB used 8.03GiB path /dev/md127
devid 2 size 29.66GiB used 8.01GiB path /dev/md126

Btrfs v3.17.2
sh-4.3#
The combos provided by this activity are illustrated in the following:
sh-4.3# blkid
/dev/sde: UUID="073f4158-1b28-4a48-a1e9-9b7840102ea8" UUID_SUB="507c657f-c864-45b4-a684-0e583aa9e26e" TYPE="btrfs"
/dev/sdd: UUID="073f4158-1b28-4a48-a1e9-9b7840102ea8" UUID_SUB="ace1834f-a628-4159-923f-bc3073366b02" TYPE="btrfs"
/dev/sdb: UUID="073f4158-1b28-4a48-a1e9-9b7840102ea8" UUID_SUB="6155a370-063d-4c5c-9e60-5f4bf97763f4" TYPE="btrfs" PTTYPE="dos"
/dev/sdc: UUID="073f4158-1b28-4a48-a1e9-9b7840102ea8" UUID_SUB="23a2979d-3be3-4f15-bf1e-269b94225b64" TYPE="btrfs"
/dev/sda: UUID="a3659970-8231-4cf8-9f5e-ed2aab05ea51" UUID_SUB="7f7cc65f-22a9-43f2-83a6-1b7633617943" TYPE="btrfs" PTTYPE="dos"
/dev/sdl: UUID="ffddaec2-6808-ff69-fe59-23706e101601" UUID_SUB="c82559af-ddb3-77db-964c-d0b91fd30c76" LABEL="localhost:124" TYPE="linux_raid_member"
/dev/md127: UUID="7ca21c5b-e887-435a-9db3-1ad13b145aea" UUID_SUB="0f2de449-6bef-4872-99b1-b806b43ec6a4" TYPE="btrfs"
/dev/sdn: UUID="1b639595-329c-763d-d35d-3eee9ce09b90" UUID_SUB="206d9fe7-aedb-e6e3-5042-544080d51f71" LABEL="localhost:125" TYPE="linux_raid_member"
/dev/md126: UUID="7ca21c5b-e887-435a-9db3-1ad13b145aea" UUID_SUB="1bc747b4-d9ac-4707-8e9c-9567ca84acb0" TYPE="btrfs"
/dev/sdq: UUID="ffddaec2-6808-ff69-fe59-23706e101601" UUID_SUB="da5525f7-7e82-b818-4ab4-2f712a5c159c" LABEL="localhost:124" TYPE="linux_raid_member"
/dev/sds: UUID="1b639595-329c-763d-d35d-3eee9ce09b90" UUID_SUB="cfb0d45e-dfc2-5e34-058b-8e9296636c97" LABEL="localhost:125" TYPE="linux_raid_member"
sh-4.3#

There are two pairs of devices designated by identical UUID's which are joined by the btrfs system.

The size of the running system is ~112GB. I assume it is raid0 for each pair.

Any of the four devices can be selected to boot but boot occurs only on /dev/sdb which has PTTYOPE=DOS notation.
The data show /dev/sda and two mdadm arrays which are loaded from pcie USB interface.

With this experimental arrangement, the system boots from any of the devices dev sda through device sde. No selection is becessary except if /dev/sda is desired. The pcie devices do not boot.

A somewhat wild experiment. I do not know how to inspect the four device btrfs setup to evaluate its real raid performance. It should be raid10 according to btrfs info but I did not include metadata in the setup of the /dev/sdb.

It might therefore be two legs of a btrfs raid0 array with both legs raid0 or just one raid0 with 4 devices involved.

Performs nicely....

EDIT: 12/11/14 ...Upgrade to linux-3.17.6-1 with no problems. Using aura from aur with success and appreciate its performance! Everything here is root only and has been for 15 years now. Merry Christmas to all who venture here!!!

EDIT: 12/13/14...Nice date sequence today! Bought a ssdIII and installed archlinux. This connects to sataII port on my machine. The read speed is shown below;sh-4.3# hdparm -t /dev/sdc

/dev/sdc:
Timing buffered disk reads: 674 MB in 3.01 seconds = 224.28 MB/sec
sh-4.3#

The filesystem is displayed in detail below;
sh-4.3# btrfs fi show
Label: none uuid: 2d0c9532-49a2-4af8-a4b4-aaee8c8cc1dd
Total devices 1 FS bytes used 4.47GiB
devid 1 size 111.79GiB used 7.04GiB path /dev/sdc

This device copied a 640MB file at a transfer speed of 229MB/s.

The max speed in statII is ~275MB/s.

The data suggests that the performance of sataIII device in sataII gives good results.

The device is a pny sataiii 128GB capacity installed with archlinux linux-3.17.6-1 in full-capacity mode btrfs.

The boot time is shown below;
sh-4.3# systemd-analyze
Startup finished in 422ms (kernel) + 3.245s (initrd) + 1.085s (userspace) = 4.754s
sh-4.3#

The device shutsdown is ~2+ seconds

EDIT: 12/15/14...Experimented with pny ssd in pcie sata adapter. As expected, the read speed reduced to the bus speed of 133MB/s. The adapter has two sata ports and may permit faster btrfs raid0 operation with two devices except for the fact that the same bus is encountered with each device. It might be a new ecperiment to try soon.

EDIT: 12/20/14...Installed qutebrowser and find it to be a fine addition to my ssd's in archlinux. It provides a "cheat sheet" for the keybd using a png display. This display is nicely adapted by phototonic to place the details on the screen even with the system set up on the internet via qutebrowser. It serves as a reference in applying the keybd shortcuts and aiding in the application of those shortcuts. I utilize qutebrowser to access multiple url's with a file list headed by the qutebrowser command. This file is utilized in xfce4 to create the default web browser.
In operation, the url's appear at top panel and enable selection of desired url at random. Zoom is provided by "+" activation.
This activity is similar to jumanji in the same web browser arrangement.

With this setup, archlinux forums and alternative mail service are directly available with a mouse click, as well as news related url's.. It fits the KISS principle very nicely.

I note the total number of views for this post is now 12539. I assume it is related to the btrfs activities I report. I enjoy experimenting with versatile system packages.

EDIT: 12/23/14...Took a detour to try a wild experiment. I utilized a maxell 16GB compact flash device connected in a usb2 adapter to install to a udf version of archlinux to the full-capacity of the device. I utilized the following to set up the device before applying the archiso install procedure;

#mkudffs --media-type=hd --blocksize=512 /dev/sdb

Then applied the archiso via usb flash device and mounted the /dev/sdb in /mnt.

Proceeded to utilize pacstrap but added intel-ucode to the mix.

After the packages were installed, utilized gebfstab attempting to generate a UUID for the device. This did not occur and the device was Labeled linuxudf and the fstab listing was for /dev/sdb....no UUID.

Proceeded with the arch-chroot /mnt and performed the necessary install items for the system...hostname,hosts,timezone...etc.

When ready to install grub, used...... pacman -S grub.

At this point, because I was utilizing the full-capacity of the device as root, I had to use the following to obtain an install....

#grub-install --force /dev/sdb (I ignored all error messages and received a done response)

I followed this with ;

#grub-mkconfig -o /boot/grub/grub.cfg ( again waited for a completion showing successful install)

Followed with;

#mkinitcpio -p linux

exited chroot and rebooted.

First reboot failed with (load kernel first).....

Retry and edited the set root= (hd0)

This permitted the system to reach log in prompt after some minutes.

Login as root.

Perform the needed ip link setup and dns setup required for my static IP's.

checked pacman with...pacman -Syy...

Installed via pacman xorg-xinit xorg-server xorg-server-utils xf86-video-intel gvfs leafpad xfce4 and others.

This completed the main effort and at the terminus of the install was able to run;

#startxfce4

This took a while but resulted in a display of xfce4 Desktop.

Subsequently, I added many packages to complete my archlinux udf filesystem which seems only to run in usb mode.AFAICT.

I am posting via the udf system at present.

I realize it cannot be upgraded due to the limitations of grub and that my /boot file will disappear if I upgrade.

I find the system response in usb mode is snappy except for long boot time and shutdown time.

I find it to be a wild off-the-wall experiment but with success....so far!!!!!

I have otter-browser installed and qutebrowser as well.

I have to remember not to upgrade!!!!!!

EDIT: 12/25/14 ....Merry Christmas to all! Received a 60GB Patriot Blaze SSD for a gift and promptly installed archlinux UDF to the device. In sata connection the device ran 224MB/s read speed.
After install of UDF via the grub --force procedure, the read speed maxed out at 160MB/s. The boot time is shown below;

sh-4.3# systemd-analyze
Startup finished in 411ms (kernel) + 3.287s (initrd) + 1.550s (userspace) = 5.249s
sh-4.3#

/dev/sdb:
Timing buffered disk reads: 476 MB in 3.00 seconds = 158.56 MB/sec
sh-4.3#

The system contains otter-browser, qutebrowser, xfce4,mpv,gparted,hdparm , feh, leafpad , alsa-utils, flashplugin,xpdf,ffcast,and phototonic. Aura-bin is also installed. Qutebrowser is enabled as web browser and ehrn activated, connects to nine urls and provides instant connect thereafter to any of them.

I realize that I cannot upgrade this drive because it has been forced with grub and will be wiped of grub boot if I do upgrade.

I post this from the installed UDF system. It isn't as fast in hdparm as btrfs but it performs snappily.

EDIT: 12/29/14....Another side issue....installed ext4 archlinux to full-capacity compact flash 15gb device and used the.... grub --force /dev/sd(x)...procedure successfully. Waiting for an upgrade to the kernel to test a recovery of boot in this device because it will be impacted by the upgrade. Now have two udf devices and an ext4 device awaiting a new kernel.

EDIT: 1/7/15....Downgraded linux kernel and rebooted successfully. Re-installed linux with ...Syu... and rebooted successfully. This indicates that an upgrade of the kernel is not likely to remove the ability to boot the full-capacity ext4 archlinux install. I note the linux-grsec has been released but linux-3.18.6-1 remains in testing. This is likely since grsec is a major re-do for security purposes and takes everyone's activity time to process. Probable that grsec will eventually be the only archlinux format. The original aim of archlinux was to keep it simple!

EDIT: 1/9/15...........As a side issue with the latest pacman4.0.2-5, I cannot run AUR buiilds --asroot. To show the extent of my system's use of AUR packages, I sow ...pacman -Qm below;

sh-4.3# pacman -Qm
aura-bin 1.3.0.2-1
cloog 0.18.1-3
dropbox 3.0.4-1
ffcast 1:2.4.1-1
isl 0.14-1
jumanji 20110811-1
mhddfs 0.1.39-1
otter-browser 0.9.04-1
phototonic 1.5-1
python-pypeg2 2.15.1-1
qutebrowser 0.1.1-3
udftools 1.0.0b3-7
xpdf 3.04-2
xrectsel 0.3.1-1
sh-4.3#

This shows all packages are from AUR.

In order to utilize builds from the AUR under the new pacman (patched), I approached Allan McRae with a request for aid. After several attempts, the following was established to run the builds as "nobody".

#alias makepkg="sudo -u nobody makepkg"

This does enable builds but is not enough to permit without manually changing permissions to read and write in the downloaded file from the AUR.

One roblem now remains....what to do with an upgrade using ...aura -Au .... for all packages in the AUR that are extant on my system? Aura-bin is not set up for user permissions.

Only one source has introduced any problem in 11 years with --asroot at my archlinux station. That is the removal of --asroot option from pacman.

EDIT: 1/13/15.....While the previous cammand allows packages from aur by "nobody with the alias, the alias also permits the following command from "aura", the aur build package; (Full-capacity drive, no partitions,root only)

#aura --builduser=nobody -A (aur package)..... (with alias present)

Latest aura is posted as aura1-git but needs an upgrade for new pacman package so will not build.

EDIT: 1/13/15...With full sized ssd's, the root is too large to utilize suspend-to-ram because the root size is larger than the ram installed. This could be corrected by using thumb drives of ..say..2GB, thus being small enough to be handled by ram. This could then be bootless reboot. However, the system would be constricted greatly. Always, there is a limit.....

EDIT: 1/15/15.................Upgraded all drives to linux-3.18.2-2X86... with no problems. This included three drives which were grub force loaded in ext4 full-capacity devices. Thus, the 16GB adata 533x device was upgraded successfully and a sandisk 3.0 16GB device also upgraded in forced ext3 grub full-capacity. The prospec 16GB udma7 device also upgraded and has btrfs installed full-capacity grub boot. The four ssd device also responded to the upgrade with no problem. Also, the patriot blaze 128GB device loaded with ext4 forced grub responded to the upgrade. So I conclude that upgrades to the kernel are not likely to cause failure of the forced grub bootloader. The boot time for the devices is typically ~5 seconds.
The sandisk device is connected into USB 2.0 port and has a 35MB/s read speed in hdparm. My mobo cannot do better than 100mhz bandwidth bus for usb devices at present.

Ray

EDIT: 1/21/15 Reinstalled archlinux btrfs into maxell 16GB udma device which had been running in udf filesys. Since it doesn't run in fstab as uuid, the use of udf is restricted to prticular sd(x) device.
Sandisk 16GB 3.0 device has a boot time of ~5 secs and a read speed of 34MB/s in usb 2.0 port. This is a surprise since most usb2.0 devices are around 10MB/s and CF devices run not much more.

A test of a san disk 16gb thumb drive with archlinux latest kernel and system packages installed revealed that , with pcie adapter to 3.0 on 100mhz bus gave 97MB/s read and 87MB/s write...impressive readings for usb...actually better than cf devices. The device does not boot from this connection but does boot from a normal 2.0 usb port. The device is not recognized as a pcie block device in pcie port.

EDIT: 1/24/15... Performed the following test of file transfer to determine use of /tmp as go between for large files...file size in test is 640.5MB....

sh-4.3# time cp /var/local1/81 /tmp && cp /tmp/81 /root/Desktop

real 0m0.942s
user 0m0.000s
sys 0m0.940s
sh-4.3#
This is 640+MB/s transfer speed using cache. The OS is 16GB compact flash device in adapter from pata IDE port. Will repeat this test after reboot.

Test repeated and reveals that the file was in cache to begin with so invalidates the result. New data show 96MB/s transfer speed....

sh-4.3# time cp /var/local1/81 /tmp && cp /tmp/81 /root/Desktop

real 0m6.670s
user 0m0.013s
sys 0m1.243s
sh-4.3#

Retest indicates a write speed of 96MB/s from the following: File is 640.5MB....

sh-4.3# mount /dev/sdb /mnt/md
sh-4.3# time cp /var/local1/81 /tmp && cp /tmp/81 /mnt/md

real 0m6.671s
user 0m0.000s
sys 0m1.260s
sh-4.3#
The limit for the bus is around 100MB/s and the speed is consistent with usb3.0 devices altho it is write speed and not read speed. Also it is real time. The cache is not involved with this event. If it was, the speed would be about 1 second of real time.

EDIT: Ran another test...cp /var/local1/81 /tmp...then ran..... mount /dev/sdb /mnt/md....

The file is now in cache as well as /tmp....

Now ran.....time cp /tmp/81 /mnt/md...results given below:

sh-4.3# time cp /tmp/81 /mnt/md

real 0m2.333s
user 0m0.007s
sys 0m0.917s
sh-4.3#

This gives the write speed of 274MB/s which is the max speed of sataII in this mobo.

EDIT: 1/27/15...Upgraded all systems to kernel linux-3.18.3-1 with no problems. This includes devices with ext4,btrfs,and mdadm raid0 with 4 devices(ssd's). One ssd is provided with btrfs. All systems boot with less than 8 secs time, the 8n secs time is for USB 3.0 devices booting from usb2 port. All others are ~5secs boot time. All systems are root only and I have 10 devices arrangements with archlinux installed. I have not experienced any btrfs fails in two years of operating in this root only mode. My philosophy is KISS and no terabyte HDD's are KISS (except for maximum data loss with failure!.) Also, multiple partitions and dual boots are not KISS and fraught with failure modes. Using aur based packages almost exclusively.

EDIT: 1/28/15...Upgraded to linux-3.18.4-1 in all systems with no problems experienced. All devices are flash based and have been for many years. All are root only OS archlinux and frequently upgraded. So much for operating as root for at least 11 years.

EDIT: 2/2/15...Upgraded all systems to linux-3.18.5-1 and all systems run normally. Boot times still at ~5 secs on compact flash devices and ssd devices. USB 3.0 devices in 2.0 ports run abiout 8 secs as before. All systems are root only and full-capacity root only drives.

EDIT: 2/8/15...Ugraded all systems to linux-3.18.6-1 with no problems. This includes sdhc usb devices and ssd devuices and compact flash devices all root only full-capacity.

EDIT: 3/13/15 ... Another friday 13th! I expect in the near future that the upgrades planned will blast holes in my systems.

I use aura-bin for most of my packages and the scuttlebutt says it will be cremated. I am not able to install shaman-git at this time due to some byte sequence failure. I have experimented further with bootloader syslinux and was successful in booting a full-capacity btrfs system. I have since had no luck as yet with a mdadm raid0 bootloaded with syslinux since btrfs is a special case for syslinux. I haven't found a defined install and configure syslinux for a mdadm raid0 pair using full-capacity devices in a bios boot. Wiki info describes a partitioned method but describes poorly how to configure syslinux IMHOP.

I haven't given up but suspect it won't perform after the upcoming major changes in Archlinux-linux core packages. So I wait and see before wasting my time.

Ray n6re

EDIT:3/15/15 I note that 1635 packages will upgrade with GCC5 and that will take 900 minutes for each of ten systems with my adsl internet setup. If it were true that each system was ide
ntical to the others, I would be thru much quicker by copying from the initial download. Have question marks for the AUR packages I have a dozen or so of. I may have to forego many systems until later if the original download is successful because I cannot be at the machine for long periods and dislike running it on the power grid and running the light bill skyward.

EDIT: 3/18/15....Upgraded to linux-3.19.2-1 to all systems with no problems. Still trying to develop a method of installing btrfs with syslinux to 2 full-capacity devices in mdadm raid0 with no partitions. Present info is given in wiki(s) for partitioned installs only.

EDIT: 3/23/15...Attempted to build a new OS via archiso as usual but failed to startxfce4. Noted several others in the forums with the same problem after recent upgrade with pacman -Syu.
Decided to wait out the problem since it seems likely to be caused by an upgrade to xorg-xinit to versio 1.3.4-2. Startx also does not perform reporting problem with xinit.

EDIT: 4-7-15...Upgraded to linux-3.19.3-1 on all systems with no problems. I assume the problem with xfce4 was caused by a defective download from a failed mirror file. Still feel apprehensive about the GCC upgrade.

EDIT: 4/9/15 Upgraded all systems to linix-3.19.3-3with no problems.

EDIT 4/29/15 Upgraded all systems to liinux-4.0.1-1 with no problems. Noted that ssd devices reduced the boot time in systemd-analyze to best time of 4.610secs. Use of devices connected to sata port with adapters increased the boot time by approx 2 secs and is due to the adapter. A new systemd item appears in blame display called ...dev-sd(x).device...which has over 3 seconds of boot time. Have to study this new entry to try to reduce boot time.

EDIT: 5/3/15...Reached a 4 second boot time with edits to mkinitcpio, journal,and removal of internet params from the boot sequence. Installed a Desktop script to initiate the fixed IP internet and enable internet accrss in two mouse clicks. Also provided a script to allow internet access to mail...thunderbird. These changes improve system access as well as faster boot. I conclude that this level of boot time is probably my hardware limit.The present in use device is a 120GB ssd has the following boot data:

sh-4.3# systemd-analyze
Startup finished in 323ms (kernel) + 3.231s (initrd) + 501ms (userspace) = 4.057s
sh-4.3#
The initrd carries the largest time load and I have not attempted to modify that parameter.

Hdparm for the device follows:
sh-4.3# hdparm -t /dev/sdd

/dev/sdd:
Timing buffered disk reads: 682 MB in 3.00 seconds = 227.27 MB/sec
sh-4.3#
The latest kernel linux-4.0.1-1-X86....introduces a new systemd parameter which examines the hardware in use and increases the boot time thereby if adapters are utilized. It is termed....dev-sd(x).device.

EDIT: 5/14/15....Upgraded to linux-4.0.2-1-X86 in all systems with no problems. Best boot time 4.151s in btrfs raid0 pair of ssd's.

EDIT: 5/18/15... Upgraded all systems to linux-4.0.3-1-x86...........No problems encountered. Still awaiting the GCC upgrade which may happen this week.

EDIT: 5/19/15...Upgraded all systems to linux-4.0.4-1-x86 with no problems. Still expecting GCC change soon.

EDIT: 5/23/15...Upgraded all systems to linux-4.0.4-2-x86...with no problems. Also included gcc5.1.0.4-x86 which installed correctly. Noted the warning for md/raid0..ext4...corruption but have no such arrangement. All of my raids are btrfs systems and should be ok with linux4.0. Up to this point, that has been true.

EDIT: 6/3/15....The best boot time I have achieved with ssd's ia as follows:

[root@n6re ~]# systemd-analyze
Startup finished in 339ms (kernel) + 3.268s (initrd) + 393ms (userspace) = 4.001s
[root@n6re ~]#

EDIT: 6/7/15 Upgraded all systems to linux-4.0.5-1-x86..with no problems.

EDIT:....6/8/15...Some data regarding the ssd pair of 60GB/120GB devices. The root directory size is reported as 168GB in df -h. The info from hdparm is interesting as well. All data follows;
sh-4.3# systemd-analyze
Startup finished in 327ms (kernel) + 3.236s (initrd) + 505ms (userspace) = 4.069s
sh-4.3#

sh-4.3# systemd-analyze
Startup finished in 327ms (kernel) + 3.236s (initrd) + 505ms (userspace) = 4.069s
sh-4.3# hdparm -I /dev/sdb

/dev/sdb:

ATA device, with non-removable media
Model Number: SATA SSD
Serial Number: DC11075307EE01586017
Firmware Revision: S9FM02.0
Transport: Serial, ATA8-AST, SATA 1.0a, SATA II Extensions, SATA Rev 2.5, SATA Rev 2.6, SATA Rev 3.0
Standards:
Supported: 10 9 8 7 6 5
Likely used: 10
Configuration:
Logical max current
cylinders 16383 16383
heads 16 16
sectors/track 63 63
--
CHS current addressable sectors: 16514064
LBA user addressable sectors: 234439535
LBA48 user addressable sectors: 234439535
Logical Sector size: 512 bytes
Physical Sector size: 512 bytes
Logical Sector-0 offset: 0 bytes
device size with M = 1024*1024: 114472 MBytes
device size with M = 1000*1000: 120033 MBytes (120 GB)
cache/buffer size = unknown
Form Factor: 2.5 inch
Nominal Media Rotation Rate: Solid State Device
Capabilities:
LBA, IORDY(can be disabled)
Queue depth: 32
Standby timer values: spec'd by Standard, no device specific minimum
R/W multiple sector transfer: Max = 16 Current = 16
Advanced power management level: disabled
DMA: mdma0 mdma1 mdma2 udma0 udma1 udma2 udma3 udma4 udma5 *udma6
Cycle time: min=120ns recommended=120ns
PIO: pio0 pio1 pio2 pio3 pio4
Cycle time: no flow control=120ns IORDY flow control=120ns
Commands/features:
Enabled Supported:
* SMART feature set
Security Mode feature set
* Power Management feature set
* Write cache
* Look-ahead
* Host Protected Area feature set
* WRITE_BUFFER command
* READ_BUFFER command
* DOWNLOAD_MICROCODE
Advanced Power Management feature set
SET_MAX security extension
* 48-bit Address feature set
* Device Configuration Overlay feature set
* Mandatory FLUSH_CACHE
* FLUSH_CACHE_EXT
* SMART error logging
* SMART self-test
* General Purpose Logging feature set
* WRITE_{DMA|MULTIPLE}_FUA_EXT
* IDLE_IMMEDIATE with UNLOAD
* WRITE_UNCORRECTABLE_EXT command
* Segmented DOWNLOAD_MICROCODE
* Gen1 signaling speed (1.5Gb/s)
* Gen2 signaling speed (3.0Gb/s)
* Gen3 signaling speed (6.0Gb/s)
* Native Command Queueing (NCQ)
* Host-initiated interface power management
* Phy event counters
* unknown 76[13]
* unknown 76[14]
DMA Setup Auto-Activate optimization
Device-initiated interface power management
* Software settings preservation
* DOWNLOAD MICROCODE DMA command
* Data Set Management TRIM supported (limit 8 blocks)
Security:
Master password revision code = 65534
supported
not enabled
not locked
not frozen
not expired: security count
supported: enhanced erase
2min for SECURITY ERASE UNIT. 60min for ENHANCED SECURITY ERASE UNIT.
Checksum: correct
sh-4.3#

sh-4.3# systemd-analyze
Startup finished in 327ms (kernel) + 3.236s (initrd) + 505ms (userspace) = 4.069s
sh-4.3# hdparm -I /dev/sdb

/dev/sdb:

ATA device, with non-removable media
Model Number: SATA SSD
Serial Number: DC11075307EE01586017
Firmware Revision: S9FM02.0
Transport: Serial, ATA8-AST, SATA 1.0a, SATA II Extensions, SATA Rev 2.5, SATA Rev 2.6, SATA Rev 3.0
Standards:
Supported: 10 9 8 7 6 5
Likely used: 10
Configuration:
Logical max current
cylinders 16383 16383
heads 16 16
sectors/track 63 63
--
CHS current addressable sectors: 16514064
LBA user addressable sectors: 234439535
LBA48 user addressable sectors: 234439535
Logical Sector size: 512 bytes
Physical Sector size: 512 bytes
Logical Sector-0 offset: 0 bytes
device size with M = 1024*1024: 114472 MBytes
device size with M = 1000*1000: 120033 MBytes (120 GB)
cache/buffer size = unknown
Form Factor: 2.5 inch
Nominal Media Rotation Rate: Solid State Device
Capabilities:
LBA, IORDY(can be disabled)
Queue depth: 32
Standby timer values: spec'd by Standard, no device specific minimum
R/W multiple sector transfer: Max = 16 Current = 16
Advanced power management level: disabled
DMA: mdma0 mdma1 mdma2 udma0 udma1 udma2 udma3 udma4 udma5 *udma6
Cycle time: min=120ns recommended=120ns
PIO: pio0 pio1 pio2 pio3 pio4
Cycle time: no flow control=120ns IORDY flow control=120ns
Commands/features:
Enabled Supported:
* SMART feature set
Security Mode feature set
* Power Management feature set
* Write cache
* Look-ahead
* Host Protected Area feature set
* WRITE_BUFFER command
* READ_BUFFER command
* DOWNLOAD_MICROCODE
Advanced Power Management feature set
SET_MAX security extension
* 48-bit Address feature set
* Device Configuration Overlay feature set
* Mandatory FLUSH_CACHE
* FLUSH_CACHE_EXT
* SMART error logging
* SMART self-test
* General Purpose Logging feature set
* WRITE_{DMA|MULTIPLE}_FUA_EXT
* IDLE_IMMEDIATE with UNLOAD
* WRITE_UNCORRECTABLE_EXT command
* Segmented DOWNLOAD_MICROCODE
* Gen1 signaling speed (1.5Gb/s)
* Gen2 signaling speed (3.0Gb/s)
* Gen3 signaling speed (6.0Gb/s)
* Native Command Queueing (NCQ)
* Host-initiated interface power management
* Phy event counters
* unknown 76[13]
* unknown 76[14]
DMA Setup Auto-Activate optimization
Device-initiated interface power management
* Software settings preservation
* DOWNLOAD MICROCODE DMA command
* Data Set Management TRIM supported (limit 8 blocks)
Security:
Master password revision code = 65534
supported
not enabled
not locked
not frozen
not expired: security count
supported: enhanced erase
2min for SECURITY ERASE UNIT. 60min for ENHANCED SECURITY ERASE UNIT.
Checksum: correct
sh-4.3# df -h
Filesystem Size Used Avail Use% Mounted on
devtmpfs 1.6G 0 1.6G 0% /dev
tmpfs 1.6G 0 1.6G 0% /dev/shm
tmpfs 1.6G 17M 1.6G 2% /run
tmpfs 1.6G 0 1.6G 0% /sys/fs/cgroup
/dev/sdb 168G 4.3G 106G 4% /
tmpfs 1.6G 20K 1.6G 1% /tmp
sh-4.3#

The devices are BLAZE ssd's and run 230mb/s in read speed on my machine. (sataII) The devices are satIII devices.

EDIT: 6/12/15...Looks like I may lose all my AUR programs soon due to AUR4 changes. This will cause a change which may not include things I need to maintain my systems.

I note that BBcode is "on" so am trying to test it now since it wasn't responding in earlier tries.......

sh-4.3# btrfs fi show
Label: none  uuid: abe5b566-8f4b-4015-9fae-9cd3d9e5f489
    Total devices 1 FS bytes used 5.69GiB
    devid    1 size 14.92GiB used 8.04GiB path /dev/sda

warning devid 2 not found already
Label: none  uuid: 7dddf155-dcac-4ee6-ab4f-a97ad822bdb8
    Total devices 1 FS bytes used 9.59GiB
    devid    1 size 14.92GiB used 12.04GiB path /dev/sde

Label: none  uuid: d90112b6-0f5b-4021-a07c-b25f0f4ad652
    Total devices 2 FS bytes used 4.09GiB
    devid    1 size 111.79GiB used 4.02GiB path /dev/sdb
    devid    2 size 55.90GiB used 4.01GiB path /dev/sdd

Label: none  uuid: f6216a2b-89ec-4092-aff6-f1833e408981
    Total devices 2 FS bytes used 3.61GiB
    devid    1 size 55.90GiB used 5.03GiB path /dev/sdc
    *** Some devices missing

Label: none  uuid: 7ca21c5b-e887-435a-9db3-1ad13b145aea
    Total devices 2 FS bytes used 13.66GiB
    devid    1 size 29.83GiB used 8.02GiB path /dev/md127
    devid    2 size 29.66GiB used 8.01GiB path /dev/md126

btrfs-progs v4.0.1
sh-4.3#

EDIT: 6/25/15...Upgraded to linux-4.0.6-1-x86 with no problems. Using dropbox for large videos for the next year. Expecting to find a new quantom-dot monitor later this summer, should the hype be true.

EDIT: 7/1/15..Upgraded all systems to linux-4.0.7-2-x86...with no problems. Still anxious for the quantom-dot monitor to show. May have a showing before August, my ninetieth.

EDIT: 7/16/15 ...Upgraded to linux4.1.2-2-x86....on all systems with no problems. Still awaiting the quantom-dot monitors promised for summer sale.

EDIT: 7/29/15 ...Upgraded to linux-4.1.3-1x86 with no problems. Still awaiting the quantom-dot hyped monitors.

EDIT:....8/6/15...Upgraded to linux-4.1.4-1-x86 with no problems. Still awaiting quantom-dot monitor for sale!

EDIT: 8/17/15....Upgraded to linux-4.1.5=1-x86 with no problems. Still await the monitor!

EDIT: 8/24/15...Upgraded to linux-4.1.6-1-x86 with no problems. Still anxious for the nano-crystal monitor(s).

EDIT: 9/15/15 Installed dhcp to time warner cable system with dg1670 router/modem. Obtained ~45Mbps download speed and ~13Mbps upload at a random time with other computers connected. Utilized windows to determine DNS not supplied by time warner techs. They don't seem to have info for linux use.
This raises my download speed almost fifty times if oerating at full speed at all times. Not likely due to cabler users at all times. I appreciate the increase in download speed since i need to finish quickly and mind my alzheimers patient. So I now say goodbye to Verizon after 14 years of adsl at 18000 ft from central. Cost is $69/mo.in So California. All systems working well and very fast!

EDIT: 9/28/15 .... Upgraded allsystems to linux-4.2.1-1-x86.... with no problems. Still awaiting the quantom-dot displays being hyped. Btrfs has performed well for the whole series of systemd installs. I am sold on its performance for full-sized devices. with grub as the bootloader.

EDIT: 10/1/15...Upgraded to linux-4.2.2-1 with no problems. Intend at this time to utilize the first post of this thread to continue my efforts with btrfs. Please look for the latest comments with one special report which is intruguing! Cu there?

Last edited by lilsirecho (2015-10-02 02:43:25)

Arch Linux

#51 2012-07-15 01:29:54

Re: Raid0 Btrfs grub boot experiments full size devices (See Edits)

#52 2012-07-15 06:27:32

Re: Raid0 Btrfs grub boot experiments full size devices (See Edits)

#53 2012-07-16 00:46:11

Re: Raid0 Btrfs grub boot experiments full size devices (See Edits)

#54 2012-08-24 03:24:17

Re: Raid0 Btrfs grub boot experiments full size devices (See Edits)

#55 2012-08-25 21:42:46

Re: Raid0 Btrfs grub boot experiments full size devices (See Edits)

#56 2012-09-05 04:41:44

Re: Raid0 Btrfs grub boot experiments full size devices (See Edits)

#57 2012-09-08 21:47:10

Re: Raid0 Btrfs grub boot experiments full size devices (See Edits)

#58 2012-09-08 21:57:24

Re: Raid0 Btrfs grub boot experiments full size devices (See Edits)

#59 2012-09-12 15:40:56

Re: Raid0 Btrfs grub boot experiments full size devices (See Edits)

#60 2012-09-19 03:10:04

Re: Raid0 Btrfs grub boot experiments full size devices (See Edits)

Board footer