You are not logged in.

#76 2013-02-18 03:00:16

cfr
Member
From: Cymru
Registered: 2011-11-27
Posts: 7,130

Re: Changes to LVM2 and udev break LVM2 on LUKS?

What hooks are specified in mkinitcpio.conf?


CLI Paste | How To Ask Questions

Arch Linux | x86_64 | GPT | EFI boot | refind | stub loader | systemd | LVM2 on LUKS
Lenovo x270 | Intel(R) Core(TM) i5-7200U CPU @ 2.50GHz | Intel Wireless 8265/8275 | US keyboard w/ Euro | 512G NVMe INTEL SSDPEKKF512G7L

Offline

#77 2013-02-18 03:06:25

dbourgeo
Member
Registered: 2012-11-26
Posts: 30

Re: Changes to LVM2 and udev break LVM2 on LUKS?

Well I have been waiting more than 10 minutes now.

I also booted an old Ubuntu copy that relies on some different lvm partitions, so it doesn't seem to be a hardware thing.

My HOOKS line was pretty standard and has both udev and lvm2. Nothing else is changed in my mkinitcpio.conf, no modules, binaries, or files are set.

HOOKS="base udev autodetect modconf block lvm2 resume filesystems keyboard fsck"

Now I am wondering about the autodetect.

Just to confirm I did it correctly, I downgraded to linux 3.7.4-1 and device-mapper 2.02.98-1 and lvm2 2.02.98-1. I made sure lvm.conf had use_lvmetad=0 and reran mkinitcpio. Exactly the same boot behaviour. My glibc was from Jan 27, so I guess that was too old to be part of the problem (glibc 2.17-3).

I also downgraded linux-api-headers, linux-docs, and linux-headers to 3.7.4-1 out of an abundance of caution. I erased the old initramfs-linux images to make sure they were being replaced properly, and searching in vmlinux-linux finds the correct version string.

I am running downgraded right now, and waiting to see if it drops to the console. Something else must need to be downgraded to get my system back to normal.

I also tried use_lvmetad=1 in the downgraded state, no luck there. Still downgraded and not working.

HMM, I won't have a chance to try it until much later, but I noted  in the new LVM docs you should have MODULES="dm_mod" set. I am pretty sure that wasn't set. I can't imagine how I got away with that before (if indeed), and hopefully that will solve things about 8 hours from now... (I use vimdiff for the pacsave files and only remember touching the lvmetad line, but maybe I goofed by saving the pacsave version somehow).

Updated to correct to  "use_lvmetad" (was "lvmetad").

Last edited by dbourgeo (2013-02-20 22:05:17)

Offline

#78 2013-02-18 16:22:33

falconindy
Developer
From: New York, USA
Registered: 2009-10-22
Posts: 4,111
Website

Re: Changes to LVM2 and udev break LVM2 on LUKS?

dbourgeo wrote:

HMM, I won't have a chance to try it until much later, but I noted  in the new LVM docs you should have MODULES="dm_mod" set.

Not sure what "new" docs these are, but they're wrong.

Offline

#79 2013-02-18 18:24:02

dbourgeo
Member
Registered: 2012-11-26
Posts: 30

Re: Changes to LVM2 and udev break LVM2 on LUKS?

Under 3.8 Configure System of https://wiki.archlinux.org/index.php/LVM . This is the Archlinux documentation, with a few updates I noted, but not for the new install media etc.

So, are they correct or not on the device mapper module (also relevant, I am trying to downgrade)?

Last edited by dbourgeo (2013-02-18 18:44:26)

Offline

#80 2013-02-18 19:26:42

alphaniner
Member
From: Ancapistan
Registered: 2010-07-12
Posts: 2,810

Re: Changes to LVM2 and udev break LVM2 on LUKS?

dbourgeo wrote:

... the Archlinux documentation ...

Remember, the wiki is community maintained; it's not official Arch documentation.

The module is added by the lvm2 hook (see /usr/lib/initcpio/install/lvm2). Note that dm_mod and dm-mod are interchangeable.


But whether the Constitution really be one thing, or another, this much is certain - that it has either authorized such a government as we have had, or has been powerless to prevent it. In either case, it is unfit to exist.
-Lysander Spooner

Offline

#81 2013-02-18 20:42:45

dbourgeo
Member
Registered: 2012-11-26
Posts: 30

Re: Changes to LVM2 and udev break LVM2 on LUKS?

You are correct alphaniner, though my quick look saw only dm-mod in the script (little update: my system seems to be missing the dm_mod symlink). I tried it anyway with mkinitcpio.conf, with no luck. Now I have to try to piece together my backups or try to downgrade the whole system (is there a recommended set of docs there?)

I am wondering what the offending package is?

Last edited by dbourgeo (2013-02-20 22:59:06)

Offline

#82 2013-02-20 00:53:57

evolutiongap
Member
Registered: 2012-03-09
Posts: 7

Re: Changes to LVM2 and udev break LVM2 on LUKS?

I have the same problem.  My system boots off LVM (no encrypt or anything fancy).  I followed the news instructions and even double-checked that the config is correct but my system won't boot because it's unable to find the volume.  After reading around I found that I can simply activate the volume in the rescue console with "lvm lvchange -ay /dev/vg-data/myvol" and then the boot can continue.  Is someone going to add the activation of the LVM back into the bootup process?  Seems like it's getting skipped or something.

MODULES="ext2 dm_mod"
HOOKS="base udev autodetect modconf block usbinput lvm2 filesystems usr fsck shutdown"

Offline

#83 2013-02-20 01:39:26

dbourgeo
Member
Registered: 2012-11-26
Posts: 30

Re: Changes to LVM2 and udev break LVM2 on LUKS?

No matter what I try, I can't get it to drop to a rescue console during the boot. I put a few echos in my grub configuration and did confirm that things hang after the initramfs.img business is completed.

I can get an old 2.6 kernel to boot and find associated LVM2 partitions. I restored my backup prior to trying to update archlinux (from the bare metal). My backup hangs in the exact same manner. Since there aren't any extra files that can possibly be left lying around by my downgrading, and the config is exactly the same as before I started this upgrade, I can only assume something else weird is going on with my system.

Right now I am cloning onto another hard drive. I suspect something might be failing in my first hard drive (I did have trouble with LVM before, but not like this, but it did appear and disappear with no apparent cause. No software updates were involved at that time. Still, this was a year ago.). Oddly, the old kernel will boot, but it lives in separate lvm2 partitions (it shares the same /boot partition with archlinux).

I have to clone over USB 2.0, so it will take  while to find out here.

Aside: I thought if you directly edited /boot/grub/grub.cfg for GRUB2 the changes should be reflected in your boot menu straight away? Anyway, I tried it from the boot menu via the "e" key, though playing with the boot stanza there is a bit of a pain. Never mind, I was mounting the wrong partition in this case. I should have seen that one immediately.

Last edited by dbourgeo (2013-02-20 22:57:35)

Offline

#84 2013-02-20 04:28:21

zdevex
Member
Registered: 2012-08-17
Posts: 24

Re: Changes to LVM2 and udev break LVM2 on LUKS?

I just wanted to throw my hat in that this solved it for me too.

global_filter = [ "r|^/dev/fd[0-9]$|", "r|^/dev/sr[0-9]$|" ]

Thanks for all the help!

-D

Offline

#85 2013-02-20 05:11:40

evolutiongap
Member
Registered: 2012-03-09
Posts: 7

Re: Changes to LVM2 and udev break LVM2 on LUKS?

Wanted to mention that the global filter did *not fix it for me.  I kept this in my lvm.conf anyways.  Still looking for a solution for the activation of the volume.

Offline

#86 2013-02-20 06:28:01

evolutiongap
Member
Registered: 2012-03-09
Posts: 7

Re: Changes to LVM2 and udev break LVM2 on LUKS?

brain0, RE: @everyone who gets dropped in a recovery shell: check udevadm info --query=all /dev/your_physical_volume and look if ID_FS_TYPE is set. Also quickly check using 'ps' what processes are running.

Thanks for the suggestion I checked if ID_FS_TYPE was set on my physical volume and it's not.  What's the solution if this is unset?  Here is the output from the udevadm command you suggested:

http://pastie.org/6234419

The process list shows lvmetad is running.

Offline

#87 2013-02-20 16:58:41

jmu
Member
From: Finland
Registered: 2012-03-18
Posts: 34

Re: Changes to LVM2 and udev break LVM2 on LUKS?

I also still suffer from the issue. All the configs and flags should be ok, but the server won't boot -
there have been some progress, but no solution

# /etc/mkinitcpio.conf
HOOKS="base udev autodetect block lvm2 filesystems keymap keyboard fsck"
MODULES=""
# /etc/lvm/lvm.conf
use_lvmetad = 1

# udevadm info (I have two disc with HW raid 1)
/dev/sda1: LVM_member
/dev/sda2: Ext4
/dev/sdb1: LVM_member
/dev/sdb2: Ext4
  1. After upgrade (boot drops into maintenance shell)

    :: Triggering uevents
    end_request: I/O error, dev fd0, sector 0
    ERROR: device '/dev/mapper/thor-root' not found
    ERROR: unable to find root device '/dev/mapper/thor-root'
  2. Set global_filter (no difference)

    # lvm.conf
    global_filter = [ "r|^/dev/fd[0-9]$|", "r|^/dev/sr[0-9]$|" ]
    
    # Still getting the error
    :: Triggering uevents
    end_request: I/O error, dev fd0, sector 0
  3. Disabled floppy drive in BIOS (did help with error, but now fsck fails)

    :: Triggering uevents
    :: performing fsck on '/dev/mapper/thor-root'
    Error reading block 91783168 (Invalid argument).
    /dev/mapper/thor-root: UNEXPECTED INCONSISTENCY; RUN fsck MANUALLY
  4. Running fsck manually (ends up in a loop)

    fsck /dev/mapper/thor-root
    Error reading block 91783168 (Invalid argument). Ignore error<y>? yes
    Force rewrite<y>? yes
    Superblock has an invalid journal (inode 8).
    Clear<y>? yes
    *** ext3 journal has been deleted - filesystem is now ext2 only ***
    Superblock has_journal flag is clear, but a journa inode is present.
    Clear<y>? yes
    The filesystem size (according to the superblock) is 183500800 blocks
    The physical size of the device is 13107200 blocks
    Either the superblock or the partition table is likely to be corrupt!
    Abort<y>? yes
    
    Error reading block 91783168 (Invalid argument). Ignore error<y>?

I'm not sure what to try next. Downgrading? Is the filesystem failure still related to the issue?
If I run the fsck from recovery disc, it completes without errors.

Update1: Added udevadm info
Update2: Added fsck from recovery info
Update3: Fixed typo in use_lvmetad

Last edited by jmu (2013-02-21 05:51:27)

Offline

#88 2013-02-20 18:52:01

dbourgeo
Member
Registered: 2012-11-26
Posts: 30

Re: Changes to LVM2 and udev break LVM2 on LUKS?

When I run blkid /dev/mapper/arch_root, it tends to report the UUID the first time, but a lot of repetition gets me a lot of failures. Someone else reported this last year but the issue was closed. I don't know if this is peculiar to my hardware somehow (as I had LVM issues before that were clearly hardware related). I wonder if the kernel can't find my UUIDs because of something similar.

I am investigating whether my hard disk and/or partition tables are involved in this, as well.  My file system checks are OK, but cfdisk reports 2 unusable areas I don't recall being there. I am cloning a new disk partition by partition now to make sure there are no issues with that.

Updated to "the kernel" from "GRUB". Oopsie, not enough sleep. Initramfs has finished loading based on a grub echo test. Well, something hangs after that echo, the last thing in my grub.cfg.

Last edited by dbourgeo (2013-02-20 22:00:22)

Offline

#89 2013-02-20 20:57:27

zdevex
Member
Registered: 2012-08-17
Posts: 24

Re: Changes to LVM2 and udev break LVM2 on LUKS?

jmu wrote:

I also still suffering from the issue. All the configs and flags should be ok, the server won't boot -
there have been some progress, but no solution

# /etc/mkinitcpio.conf
HOOKS="base udev autodetect block lvm2 filesystems keymap keyboard fsck"
MODULES=""
# /etc/lvm/lvm.conf
lvmetad = 1

While this is probably not your problem, when you say

lvmetad=1

For you /etc/lvm/lvm.conf, you do actually mean,

use_lvmetad=1

right?

I've noticed that "lvmetad=1" has been thrown around this tread as an abbreviation, but it should definitely be "use_lvmetad=1" in lvm.conf

Offline

#90 2013-02-20 21:11:02

jmu
Member
From: Finland
Registered: 2012-03-18
Posts: 34

Re: Changes to LVM2 and udev break LVM2 on LUKS?

zdevex wrote:

While this is probably not your problem, when you say

lvmetad=1

For you /etc/lvm/lvm.conf, you do actually mean,

use_lvmetad=1

right?

I've noticed that "lvmetad=1" has been thrown around this tread as an abbreviation, but it should definitely be "use_lvmetad=1" in lvm.conf

Right. The output is being manually copied, so there can be these little typos in text. But thanks for bringing it up, I'll update the post. Meanwhile, I tried downgrading some packages (linux, lvm2, device-mapper) and re-running the mkinitcpio (this time, use_lvmetad = 0), but no success there.

Last edited by jmu (2013-02-20 21:12:03)

Offline

#91 2013-02-21 17:11:29

brain0
Developer
From: Aachen - Germany
Registered: 2005-01-03
Posts: 1,382

Re: Changes to LVM2 and udev break LVM2 on LUKS?

Quick note: Your output from the boot process is not helpful when you boot with the 'quiet' option.

I also did ask for additional information on an earlier page of this thread, which so far NOBODY provided. It doesn't help if anyone repeats the same over and over and nobody actually provides the information that is asked.

At this point, I am pretty sure that one problem is inside lvm's 'pvscan' command which blocks the completion of a udev rule without a good reason. However, I still don't have confirmation on that, and it seems this is not the only problem.

Another problem is that lvmetad hangs at startup before forking to the background. I can not begin to find out why this happens, and cannot reproduce it.

Offline

#92 2013-02-21 21:35:55

evolutiongap
Member
Registered: 2012-03-09
Posts: 7

Re: Changes to LVM2 and udev break LVM2 on LUKS?

brain0, what does it mean if  ID_FS_TYPE is not set on my pv?  This pv I'm using was created a long time ago with lvm tools that apparently did not set this flag.  How can I fix this?

http://pastie.org/6234419

Offline

#93 2013-02-22 02:07:09

dbourgeo
Member
Registered: 2012-11-26
Posts: 30

Re: Changes to LVM2 and udev break LVM2 on LUKS?

If you are referring to me brain0 that output was typed in by hand, it was from the recovery grub entry, which doesn't have the "quiet" parameter.

I had everything cloned to a new drive and still had the same behaviour. I reupgraded that drive by chrooting. Still the same, everything hangs. Then I thought that my system might be hanging because of an out of date UUID for the resume kernel parameter. That seemed to get things to the point where it can now drop to a shell in initramfs. Of course, when I tried changing the UUID back to the bad one, my system did not go back to hanging again. Thankfully, I can get to a rescue console and run "lvm vgscan; lvm vgchange -ay" and that resumes the boot.

I don't know if my monkeying around with partitions and/or the upgrading and downgrading did anything (maybe the timing of activities in the initramfs stage). It is kind of strange all this started after upgrading and then half disappeared when I corrected the resume kernel parameter.

Hopefully it will continue half booting and I can send some information. Right now I am solidifying my backups because I still suspect some underlying hardware problem.

Offline

#94 2013-02-22 19:11:42

dbourgeo
Member
Registered: 2012-11-26
Posts: 30

Re: Changes to LVM2 and udev break LVM2 on LUKS?

Here is what my console shows from grub's recovery entry:

usual stuff...
running hook [resume]
Waiting 10 seconds for device /dev/disk/by-uuid/xxxxx
ERROR: resume: hibernation device 'xxx' not found
Waiting 10 seconds for device /dev/mapper/myvol-arch--root ...
ERROR: device 'xxx-root' not found. Skipping fsck.
ERROR: Unable to find root device 'xxx-root'.
You are being dropped to a recovery shell
etc.

Then I run udevadm info --query=all /dev/sda4. I have ID_FS_VERSION=LVM2 0001 and ID_FS_TYPE=LVM2_member. I have an MBR system, btw, with all my LVM2 partitions on the last partition sda4.

Running ps shows udevd  --daemon --resolve-names=never and lvmetad are running with VSZ numbers. Everybody else except init, sh -i, and ps are 0, which I expect is fully normal for this.

The  UUID E: ID_FS_UUID=aoUGTb-mrOE-dfIj-2nqs-Z9Ji-pCXw-tc1muz corresponds with my LVM2 physical partition, which is /dev/sda4.

Also the global_filter fix does nothing for me!

I will post the full output below:

P: /devices/pci0000:00/0000:00:08.0/ata1/host0/target0:0:0/0:0:0:0/block/sda/sda4
N: sda4
S: disk/by-id/ata-ST2000DM001-9YN164_Z1E0AZM4-part4
S: disk/by-id/wwn-0x5000c5003fc05368-part4
E: DEVLINKS=/dev/disk/by-id/ata-ST2000DM001-9YN164_Z1E0AZM4-part4 /dev/disk/by-id/wwn-0x5000c5003fc05368-part4
E: DEVNAME=/dev/sda4
E: DEVPATH=/devices/pci0000:00/0000:00:08.0/ata1/host0/target0:0:0/0:0:0:0/block/sda/sda4
E: DEVTYPE=partition
E: ID_ATA=1
E: ID_ATA_DOWNLOAD_MICROCODE=1
E: ID_ATA_FEATURE_SET_APM=1
E: ID_ATA_FEATURE_SET_APM_CURRENT_VALUE=254
E: ID_ATA_FEATURE_SET_APM_ENABLED=1
E: ID_ATA_FEATURE_SET_HPA=1
E: ID_ATA_FEATURE_SET_HPA_ENABLED=1
E: ID_ATA_FEATURE_SET_PM=1
E: ID_ATA_FEATURE_SET_PM_ENABLED=1
E: ID_ATA_FEATURE_SET_SECURITY=1
E: ID_ATA_FEATURE_SET_SECURITY_ENABLED=0
E: ID_ATA_FEATURE_SET_SECURITY_ENHANCED_ERASE_UNIT_MIN=222
E: ID_ATA_FEATURE_SET_SECURITY_ERASE_UNIT_MIN=222
E: ID_ATA_FEATURE_SET_SECURITY_FROZEN=1
E: ID_ATA_FEATURE_SET_SMART=1
E: ID_ATA_FEATURE_SET_SMART_ENABLED=1
E: ID_ATA_ROTATION_RATE_RPM=7200
E: ID_ATA_SATA=1
E: ID_ATA_SATA_SIGNAL_RATE_GEN1=1
E: ID_ATA_SATA_SIGNAL_RATE_GEN2=1
E: ID_ATA_WRITE_CACHE=1
E: ID_ATA_WRITE_CACHE_ENABLED=1
E: ID_BUS=ata
E: ID_FS_TYPE=LVM2_member
E: ID_FS_USAGE=raid
E: ID_FS_UUID=aoUGTb-mrOE-dfIj-2nqs-Z9Ji-pCXw-tc1muz
E: ID_FS_UUID_ENC=aoUGTb-mrOE-dfIj-2nqs-Z9Ji-pCXw-tc1muz
E: ID_FS_VERSION=LVM2 001
E: ID_MODEL=ST2000DM001-9YN164
E: ID_MODEL_ENC=ST2000DM001-9YN164\x20\x20\x20\x20\x20\x20\x20\x20\x20\x20\x20\x20\x20\x20\x20\x20\x20\x20\x20\x20\x20\x20
E: ID_PART_ENTRY_DISK=8:0
E: ID_PART_ENTRY_NUMBER=4
E: ID_PART_ENTRY_OFFSET=252274688
E: ID_PART_ENTRY_SCHEME=dos
E: ID_PART_ENTRY_SIZE=3654754480
E: ID_PART_ENTRY_TYPE=0x8e
E: ID_PART_TABLE_TYPE=dos
E: ID_REVISION=CC4C
E: ID_SERIAL=ST2000DM001-9YN164_Z1E0AZM4
E: ID_SERIAL_SHORT=Z1E0AZM4
E: ID_TYPE=disk
E: ID_WWN=0x5000c5003fc05368
E: ID_WWN_WITH_EXTENSION=0x5000c5003fc05368
E: MAJOR=8
E: MINOR=4
E: SUBSYSTEM=block
E: USEC_INITIALIZED=15170

  PID USER       VSZ STAT COMMAND
    1 0         4536 S    {init} /usr/bin/ash /init single
    2 0            0 SW   [kthreadd]
    3 0            0 SW   [ksoftirqd/0]
    5 0            0 SW<  [kworker/0:0H]
    6 0            0 SW   [kworker/u:0]
    7 0            0 SW<  [kworker/u:0H]
    8 0            0 SW   [migration/0]
    9 0            0 SW   [rcu_preempt]
   10 0            0 SW   [rcu_bh]
   11 0            0 SW   [rcu_sched]
   12 0            0 SW   [watchdog/0]
   13 0            0 SW   [watchdog/1]
   14 0            0 SW   [ksoftirqd/1]
   15 0            0 SW   [migration/1]
   16 0            0 SW   [kworker/1:0]
   17 0            0 SW<  [kworker/1:0H]
   18 0            0 SW<  [cpuset]
   19 0            0 SW<  [khelper]
   20 0            0 SW   [kdevtmpfs]
   21 0            0 SW<  [netns]
   22 0            0 SW   [bdi-default]
   23 0            0 SW<  [kblockd]
   24 0            0 SW   [kworker/0:1]
   25 0            0 SW   [kworker/1:1]
   26 0            0 SW   [khungtaskd]
   27 0            0 SW   [kswapd0]
   28 0            0 SWN  [ksmd]
   29 0            0 SWN  [khugepaged]
   30 0            0 SW   [fsnotify_mark]
   31 0            0 SW<  [crypto]
   35 0            0 SW<  [kthrotld]
   37 0            0 SW<  [deferwq]
   38 0            0 SW   [kworker/u:1]
   45 0        22316 S    udevd --daemon --resolve-names=never
   48 0        17008 S    lvmetad
  118 0            0 SW<  [ata_sff]
  119 0            0 SW   [khubd]
  120 0            0 SW<  [firewire]
  121 0            0 SW   [scsi_eh_0]
  122 0            0 SW   [scsi_eh_1]
  125 0            0 SW   [scsi_eh_2]
  126 0            0 SW   [scsi_eh_3]
  130 0            0 SW   [scsi_eh_4]
  131 0            0 SW   [usb-storage]
  132 0            0 SW   [kworker/1:2]
  136 0            0 SW<  [kworker/1:1H]
  144 0            0 SW<  [kworker/0:1H]
  145 0            0 SW   [kworker/0:3]
  150 0            0 SW   [kworker/1:3]
  172 0         4536 S    sh -i
  180 0         4536 R    ps

Last edited by dbourgeo (2013-02-22 20:13:24)

Offline

#95 2013-02-23 03:53:28

drigdrog
Member
Registered: 2010-01-12
Posts: 4

Re: Changes to LVM2 and udev break LVM2 on LUKS?

I was running into this problem - and managed to find a partial work around -- basically I downgraded lvm2 to 2.02.98-1 - which was the most recent version in my pacman cache. And then I re-ran mkinitcpio -p linux. I also needed to manually edit /usr/lib/initcpio/install/lvm2  - since the path to dmsetup was wrong.

With that my system partially booted - atleast to the point where the lvm volumes were activated and available - but for some reason not mounted. I manually mounted them (mount -a) and then systemd was able to slowly finish the boot.

As an additional point - I noticed that lvm was crashing/seg faulting in the initramfs rescue/fallback shell - when I ran "lvm2 vgscan" or some other commands it would fail.

I wonder if there is some kind of problem between lvm2 and busybox. Or maybe the new version of lvm2 has a bug...

Last edited by drigdrog (2013-02-23 03:54:01)

Offline

#96 2013-02-23 11:47:43

brain0
Developer
From: Aachen - Germany
Registered: 2005-01-03
Posts: 1,382

Re: Changes to LVM2 and udev break LVM2 on LUKS?

evolutiongap wrote:

brain0, what does it mean if  ID_FS_TYPE is not set on my pv?  This pv I'm using was created a long time ago with lvm tools that apparently did not set this flag.  How can I fix this?

http://pastie.org/6234419

ID_FS_TYPE is a property set by udev - without it, we cannot automatically detect which devices to use for LVM assembly. Are you sure sdb is your PV (as in, the while disk, not a partition)? As far as I can see, this should work fine, but I am not 100% sure right now.

I was actually referring to a later post: Once you get dropped into the recovery shell, QUICKLY run the ps command and see which processes are running (only the ones without [square brackets]). In particular, is there a 'pvscan' process still running?

Offline

#97 2013-02-23 21:55:45

dbourgeo
Member
Registered: 2012-11-26
Posts: 30

Re: Changes to LVM2 and udev break LVM2 on LUKS?

Nope, no pvscan happening at all! I had it in the buffer, just needed to hit Enter. Then I tried it with Enter in the buffer <doh>.

Addition: Possibly relevant is when I run lvchange after being dropped to the console while booting, I get the notice that lvmetad is running but disabled.

Another addition: Realised another message amongst the lvm volume notifications (when it finds the volumes) is the lvmeventd can't be stat'd. It was expected to be found in /usr/lib somewhere.

Last edited by dbourgeo (2013-02-25 03:10:31)

Offline

#98 2013-02-24 16:05:42

Fedes
Member
From: Buenos Aires, Argentina
Registered: 2012-10-23
Posts: 36

Re: Changes to LVM2 and udev break LVM2 on LUKS?

My system didn't boot as well after pacman -Syu and following the changes to lvm (mkinitcpio -p linux twice, and checked the config for lvmetad). I got a recovery console and could continue the boot process after running these commands:

lvmetad
lvm vgmknodes
udevadm trigger

lvmetad  was not running when I first checked all processes, so I had to manually start it. I don't have encyption on my system, LVM is for all partitions except boot.

EDIT; After I booted, I rerun mkinitcpio, and this time it took a long time, it could mount root and usr lvm partitions, but then failed mounting /var and got to rescue mode... this is getting annoying.
EDIT 2 : now I cannot even boot trying the same solution. Either it does not work, or gives a kernel panic.
LAST EDIT: Now i could boot normally, after removing a cd from the tray (not sure if it could be related). I checked journalctl, and saw that lvmetad service failed before with core dumps, or segfaults.

Last edited by Fedes (2013-02-24 16:32:26)

Offline

#99 2013-02-27 21:06:18

Sr. Ssruno
Member
From: Asunción, Paraguay
Registered: 2011-04-27
Posts: 12

Re: Changes to LVM2 and udev break LVM2 on LUKS?

At the rescue shell that prompts. Applying "lvm vgscan" and "lvm vgchange -ay" boots up. But is not permanent, some clues to make it work? What about putting lvm2 at the MODULES=() array?

dbourgeo wrote:

If you are referring to me brain0 that output was typed in by hand, it was from the recovery grub entry, which doesn't have the "quiet" parameter.

I had everything cloned to a new drive and still had the same behaviour. I reupgraded that drive by chrooting. Still the same, everything hangs. Then I thought that my system might be hanging because of an out of date UUID for the resume kernel parameter. That seemed to get things to the point where it can now drop to a shell in initramfs. Of course, when I tried changing the UUID back to the bad one, my system did not go back to hanging again. Thankfully, I can get to a rescue console and run "lvm vgscan; lvm vgchange -ay" and that resumes the boot.

I don't know if my monkeying around with partitions and/or the upgrading and downgrading did anything (maybe the timing of activities in the initramfs stage). It is kind of strange all this started after upgrading and then half disappeared when I corrected the resume kernel parameter.

Hopefully it will continue half booting and I can send some information. Right now I am solidifying my backups because I still suspect some underlying hardware problem.

Offline

#100 2013-02-27 21:31:43

jmu
Member
From: Finland
Registered: 2012-03-18
Posts: 34

Re: Changes to LVM2 and udev break LVM2 on LUKS?

As mentioned earlier (https://bbs.archlinux.org/viewtopic.php … 5#p1234555) I had partial success by disabling floppy disc in BIOS, but then end up with following / similar error:

:: performing fsck on '/dev/mapper/thor-root'
Error reading block 91783168 (invalid argument)
*** FILESYSTEM CHECK FAILED ***
sh: can't access tty: job control turned off
[rootfs /]# [    2.083392] usb 3-3: string descriptor 0 malformed (err = -61),
efaulting to 0x0409
[   2.178464] hid-generic 0003:1852:DB96.0003: item fetching failed at offset 36336826

Now, running `fsck -f` from rescue cd completes without errors. I even managed to boot successfully couple of times after running the fsck from cd and rebooting again, but not anymore. It seems to fail randomly(?): usually ending up to error shown above, but sometimes services startup ~normally. In latter case, the eth0 is missing for some reason, so it is not really usable for me.

As for the input for developers, what more can I provide to help? Really would not like to re-install the whole server because of this sad  Then again, current setup uses motherboard raid, while linux sw raid would have been better option I suppose. I'm tempted to disable fsck hook, but that may not be smart move, either.

- lvmetad process is running
- ID_FS_TYPE is set with value LVM2_member

Last edited by jmu (2013-02-27 21:32:56)

Offline

Board footer

Powered by FluxBB