You are not logged in.
according to namsys it has less fragmentation. I can't imagine it being worse...
Offline
Machiavelli wrote:ext3 for the root partition, reiser4 for the data disks.
I'm a little scared about what's gonna happen when I have to run a kernel which doesn't support reiser4... so I should be going back to ext3.
ext2 for /boot.
I think I heard a rumor that resier4 was going to be supported in kernel 2.6.16, dunno how true it is. It'll be supported soon enough though I'm sure.
Im feeling deja vu.
I swear that I heard someone say that early last year.....
Offline
Penguin wrote:I think I heard a rumor that resier4 was going to be supported in kernel 2.6.16, dunno how true it is. It'll be supported soon enough though I'm sure.
Im feeling deja vu.
I swear that I heard someone say that early last year.....
Hans Reiser told in an interview that "they had just missed it for 2.6.14, and were pushing for 2.6.15"...
Offline
I am an xfs / jfs fan.
Frumpus ♥ addict
[mu'.krum.pus], [frum.pus]
Offline
XFS is very very fast, but pacman does not like it at all, although it seems quite as fast as ReiserFS with ABS. It also has very fast read speeds, and doesn't work your HDD that much... except with pacman.
Re Reiser4 fragmentation problems, Hans Reiser admitted that it was a problem. When (and if!) Reiser4 is included in the Linux kernel, the Reiser4 utils package will include an online repacker a la xfs_fsr.
Offline
Hans has been saying it's going to be in the "next version of the kernel" for quite some time. It seems that the only way he thinks to accomplish this is to tell the kernel devs that they are fools to establish standards and that the kernel will suck until Reiser4 comes in. Oh, and ext3+dir_index STILL runs faster than Reiser4, plus it doesn't slow down boot-up. I think the only reason he still cares is the amount of investment he supposedly has made into the project.
Oh, and I encountered my first instability issues with Reiser4 on my last install of Arch. I'm sticking with ext3+dir_index.
Offline
Heya,
I like jfs and xfs. I'm using it on my laptop at the moment. The speed of pacman is bearable using jfs. Can be because I have 1 GB of memory in my laptop. Both are really nice filesystems in my opinion. Just don't forget to install the jfs_tools during install or maybe at first startup! Jfs needs it to replay its journal at startup, while xfs seems to function fine without them, but I installed them also.
Both are fast and safe filesystems I think.
Xfs can delete very fast in my opinion and it seems to be a safe filesystem. I use it for my /(including boot)-directory and my home-directory. I have used it (and am still using it) for quite some time on my desktop (128 MB memory) for the media-partition, where it performed nicely. I'm using jfs for /var, /opt and /usr. I'm just starting to use jfs, but it also seems a nice filesystem.
greetings,
Michel
Offline
Indexed ext3 for speed, JFS for efficiency and scalability.
Offline
ReiserFS 3.6. I like having dynamic inode creation 8)
Offline
Currently using JFS for /, /home and reiser3.6 for /var. Thinking of giving ext3+dir_index a shot now that everyone is saying how good it is
Offline
I tried ext3+dir. It is fast, but I noticed fragmentation was horrible. I run jfs for my partitions, and on the long run, it is my choice - in case of speed and stabillity.
Offline
Reiserfs 3.6
Not brave to try reiserfs4, but 3.6 never lets me down.
Offline
I actually use Reiser 3, but haven't noticed any significant performance increases between that and ext 3. I'm do for some arch installations soon. I will probably stick with ext3
Offline
Hans has been saying it's going to be in the "next version of the kernel" for quite some time. It seems that the only way he thinks to accomplish this is to tell the kernel devs that they are fools to establish standards and that the kernel will suck until Reiser4 comes in. Oh, and ext3+dir_index STILL runs faster than Reiser4, plus it doesn't slow down boot-up. I think the only reason he still cares is the amount of investment he supposedly has made into the project.
Oh, and I encountered my first instability issues with Reiser4 on my last install of Arch. I'm sticking with ext3+dir_index.
This battle is still going on? As much as i love Reiser FS, i would love for someone to punch Hans in the face, just once. Not too hard, just like a sucker punch. That said I would like to see Reiser 4 in the kernel fairly soon.
Offline
Maybe not... From the looks of benchmarks, Reiser4 doesn't really perform much better than ReiserFS 3.6 - quite a bit slower for a lot of things actually. Could be more a problem with the driver than the filesystem itself though, I mean look how much unindexed ext3 filesystems have improved...
Offline
I've been running my laptop on reiser3 exclusively for a couple of years, but as I'm about to build a new desktop, I'm kind of undecided between ext3 and reiser. The new box will run 24/7 and will prolly grow to become a part time server as well as desktop
With reiser3, I have the feeling that the disk never really gets to spin down for a long time. Even when the computer is left idle, it sounds like the disk is periodically accessed for a split second. Can anybody tell me if ext3 is better than reiser at letting the disk rest when nothing special is going on?
Have I understood other posters right in that ext3 handles many small files badly?
All of your mips are belong to us!!
Offline
Yes, it is still not that good for pacman.
Give JFS a try... I'm using it right now, and it's pretty impressive.
BTW, I have a question... Does dir_index on ext3 increase the rate of fragmentation, or is ext3 just like that? It seems to me that ext3 filesystems get kind of slow if they aren't fscked occasionally. :?
Offline
bogomipz: To minimize disk access when the system is idle, edit syslogng.conf in /etc, and direct all messages to the console. My laptop's harddrive doesn't spin up at all when I'm not using it.
Gullible Jones: Fsck checks and repairs damaged file systems, it doesn't defrag them. Thats why journaled file systems can recover more quickly from file system errors than non journaled ones - the journal pinpoints the sectors that fsck needs to repair. Non journaled ones must scan the entire platter to find inconsistancies.
Offline
Gullible Jones: Fsck checks and repairs damaged file systems, it doesn't defrag them. Thats why journaled file systems can recover more quickly from file system errors than non journaled ones - the journal pinpoints the sectors that fsck needs to repair. Non journaled ones must scan the entire platter to find inconsistancies.
Actually, from 'man e2fsck':
-D Optimize directories in filesystem. This option causes e2fsck
to try to optimize all directories, either by reindexing them if
the filesystem supports directory indexing, or by sorting and
compressing directories for smaller directories, or for filesys-
tems using traditional linear directories.
Offline
Yeah, JFS RoLzZz, is very fast and with it, I dont notice noticeable fragmentation in months !
Arch Linux Powered
Offline
I use: In Linux : JFS (The best)
and as of now I use UFS2 since I'm a FreeBSD user myself as of 3 weeks ago.
Now full support for UFS2 is something I would very much like to see in Linux... From what I've heard, UFS2 is fast and extremely scalable (moreso than XFS).
Offline
I use ReiserFS. I got tired of the system checking the integrity of the partition (or whatever it's called) during start up.
This signature just crossed the line.
Offline
I think people fuss too much over their filesystems...........
The tried and true ext3 works fine.
Ironic I say this when I distribute the popular R4 inclusive ArchCK.
iphitus
Offline
That's interesting... I find that choice of filesystem can make a surprisingly huge difference in the performance of certain applications. Just look at XFS with pacman...
Offline