Remount a LVM Vol after a system crash

Hello,
A few days ago my system crashes and i’ve used LVM v 0.8 with one volume group(server), two physical volumes(hda6, hdb6) and one logical volume(dev/server/ftp)
Now i’ve installed Suse 7.2 with LVM v 0.9 Beta7, but i can’t mount the logical volume.
The error message is:

# mount /dev/server/ftp /mnt
mount: wrong fs type, bad option, bad superblock on /dev/server/ftp,
       or too many mounted file systems

I get the following outputs from my system:

# vgscan
vgscan -- reading all physical volumes (this may take a while...)
vgscan -- found active volume group "server"
vgscan -- "/etc/lvmtab" and "/etc/lvmtab.d" successfully created
vgscan -- WARNING: This program does not do a VGDA backup of your volume
group

# pvscan
pvscan -- reading all physical volumes (this may take a while...)
pvscan -- ACTIVE   PV "/dev/hda6" of VG "server" [37.78 GB / 0 free]
pvscan -- ACTIVE   PV "/dev/hdb5" of VG "server" [41.92 GB / 0 free]
pvscan -- total: 2 [79.71 GB] / in use: 2 [79.71 GB] / in no VG: 0 [0]

# pvdisplay /dev/hda6
--- Physical volume ---
PV Name               /dev/hda6
VG Name               server
PV Size               37.78 GB / NOT usable 3.75 MB [LVM: 267 KB]
PV#                   1
PV Status             available
Allocatable           yes (but full)
Cur LV                1
PE Size (KByte)       4096
Total PE              9671
Free PE               0
Allocated PE          9671
PV UUID               /dev/h-da6

# pvdisplay /dev/hdb5
--- Physical volume ---
PV Name               /dev/hdb5
VG Name               server
PV Size               41.93 GB / NOT usable 3.45 MB [LVM: 271 KB]
PV#                   2
PV Status             available
Allocatable           yes (but full)
Cur LV                1
PE Size (KByte)       4096
Total PE              10732
Free PE               0
Allocated PE          10732
PV UUID               /dev/h-db5

# vgchange -ay
vgchange -- volume group "server" already active

# lvchange -ay /dev/server/ftp
lvchange -- logical volume "/dev/server/ftp" is already active
lvchange -- availability of logical volume "/dev/server/ftp" doesn't
have to be changed
lvchange -- logical volume "/dev/server/ftp" didn't have to be changed

I’ve tried to follow the instructions of the mailing list article, i’ve found at http://www.geocrawler.com/archives/3/3500/2001/4/0/5655568/
But it doesn’t work, After vgchange –ay I get an error message, too, like

vgcfgrestore -- ERROR: different structure size stored in "/etc/lvmconf/export.conf"
than expected in file vg_cfgrestore.c [line

"export"

I hope you can help me,
Thanks
Christoph

Christoph,

I assume that you are suffering from a PE missalignent flaw which was in 0.9 Beta 7.

Please update to the recent LVM version available at www.sistina.com. Follow the information in INSTALL and PATCHES/README to get your SuSE system updated. You need to have the kernel source package and the gcc packages installed in order to be able to do this.

Regards,
Heinz -- The LVM Guy --

*** Software bugs are stupid.
Nevertheless it needs not so stupid people to solve them ***

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

Heinz Mauelshagen Sistina Software Inc.
Senior Consultant/Developer Am Sonnenhang 11
56242 Marienrachdorf Germany
[email protected] +49 2626 141200

                                                       FAX 924446

=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-

hi everyone, I've read a number of dialogues on the archives concerning the errors below which I'm getting but have not found any answers:

I'm running RedHat 7.0 (2.4.9 kernel + LVM 1.0.1rc1) on a little PIII box and I'm using the ide controllers built into the motherboard (drivers built into the kernel - not modules). I have CONFIG_IDEDISK_MULTI_MODE set to y in the .config for the kernel. The drive is an IBM-DTLA-307030 (as reported by /proc/ide) and is brand new.

here are the errors:

Oct 17 02:00:34 beowulf kernel: hdc: read_intr: error=0x04 {
DriveStatusError }
Oct 17 02:00:34 beowulf kernel: hdc: read_intr: status=0x59 { DriveReady
SeekComplete DataRequest Error }
Oct 17 02:00:34 beowulf kernel: hdc: read_intr: error=0x04 {
DriveStatusError }
Oct 17 02:00:34 beowulf kernel: ide1: reset: success

I've been moving data in and out of the drive and everything seems ok but I get lots of the above errors which leaves me with a queasy feeling... anyone know what the deal is with this?

1k thx - e

I'm afraid they look a lot like hardware errors to me. It might just be that it doesn't like MULTI_MODE though. I'm not an IDE expert.

patrick

Patrick,

that it

I turned MULTI_MODE on at a suggestion aimed to fix these very problems... they certainly are hardware errors but as we know anecdotally from other postings, the hardware can be perfectly ok and the errors still persist. as my drive is brand new and I had no such errors before installing a PV on it, it seems reasonable to think it's not the drive.

is my data at risk of corruption? I've slowly been checking the integrity of files but I'm concerned about backing up bad data!

-----Original Message-----
From: [email protected] [mailto:[email protected]] On
Behalf Of Patrick Caulfield
Sent: Wednesday, October 17, 2001 7:30 AM
To: [email protected]
Subject: Re: [linux-lvm] read_intr errors

the

built

reported

I

I'm afraid they look a lot like hardware errors to me. It might just be that it
doesn't like MULTI_MODE though. I'm not an IDE expert.

patrick

_______________________________________________
linux-lvm mailing list
[email protected]
http://lists.sistina.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html

Just because the drive is new doesn't mean it's not bust(!).

I would do as someone suggested and run badblocks on the disk and if it gives errors then take it back. If not then it may be related to some reports we saw a while ago about LVM reading past the end of a SCSI disk - you can easily verify this by making a filesystem on the PV rather that using LVM (If you're happy with your backups that is...).

patrick

Patrick,

sorry for the delay, it takes 12 hours to back up this guy and another 12 to verify... and I had problems so I had to do it several times.

so if I get you correctly, I should be able to do badblock check (instead of a mke2fs with -c option as has been suggested) and get an evaluation for whether this drive has problems equally well, no?

here's what I did:

# df -k
Filesystem           1k-blocks      Used Available Use% Mounted on
/dev/hda1             38456308   2670876  33831932   8% /
/dev/LVM/mp3z         29540436  13308716  14731152  48% /var/mp3z

# umount /dev/LVM/mp3z
# vgchange -a n
vgchange -- volume group "LVM" successfully deactivated

root@beowulf:/root # badblocks -s /dev/hdc 29540436 Checking for bad blocks (read-only test): done

as you can see, no errors. Now, having said that, I've come to realise that the errors I get happen when accessing my tape drive (which is scsi) - at least I think that's what's happening... weird. do you then think this is related to the problems others have had? and if so, was there a fix?

- e r i c k

-----Original Message-----
From: [email protected] [mailto:[email protected]] On
Behalf Of Patrick Caulfield
Sent: Thursday, October 18, 2001 12:42 AM
To: [email protected]
Subject: Re: [linux-lvm] read_intr errors

as

it,

Just because the drive is new doesn't mean it's not bust(!).

I would do as someone suggested and run badblocks on the disk and if it gives
errors then take it back. If not then it may be related to some reports we saw
a while ago about LVM reading past the end of a SCSI disk - you can easily verify this by making a filesystem on the PV rather that using LVM (If you're
happy with your backups that is...).

patrick

_______________________________________________
linux-lvm mailing list
[email protected]
http://lists.sistina.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html

Well that looks OK

The thread about SCSI errors just stopped so I don't know what happened there. There was some thought that maybe LVM was trying to access beyond the end of the disk so you could check to see if the LV is using all the space on /dev/hdc and back it off a little so see if that helps. It may also be worth upgrading to LVM 1.0.1rc4.

If it is the SCSI tape drive interfering then that's really odd because SCSI shouldnot interfere with IDE unles you have some interrupt clashes or perhaps bad cabling causing signal noise.

Not very conclusive I'm afraid, but I did say I wasn't an IDE expert!

patrick

Patrick,

looking over the man page on lvreduce it states:

the above warning is not conclusive as to whether data loss occurs only when the LV is full i.e. it has no choice but to lose data, of if data loss may occur regardless i.e. there's no smarts in lvreduce to move data around...

can you clarify this point and perhaps reword the above for a next release of that man page?

anyhow, I understand what you're suggesting... however it's not clear to me what "a little" means... would 1K do? 1M?

here's what I've got:

# lvdisplay /dev/LVM/mp3z
--- Logical volume ---
LV Name                /dev/LVM/mp3z
VG Name                LVM
LV Write Access        read/write
LV Status              available
LV #                   1
# open                 1
LV Size                28.62 GB
Current LE             7327
Allocated LE           7327
Allocation             next free
Read ahead sectors     120
Block device           58:0

# pvdisplay /dev/hdc
--- Physical volume ---
PV Name               /dev/hdc
VG Name               LVM
PV Size               28.63 GB / NOT usable 6.69 MB [LVM: 152.00 KB]
PV#                   1
PV Status             NOT available
Allocatable           yes (but full)
Cur LV                1
PE Size (KByte)       4096
Total PE              7327
Free PE               0
Allocated PE          7327
PV UUID               vY9apM-RXf6-DtiS-m4LI-u7qT-iNjm-9SYnMz

with the above, would I do:

# lvreduce -l -1 /dev/LVM/mp3z

to test this theory?

1k thx - e

-----Original Message-----
From: [email protected] [mailto:[email protected]] On
Behalf Of Patrick Caulfield
Sent: Wednesday, October 24, 2001 2:14 AM
To: [email protected]
Subject: Re: [linux-lvm] read_intr errors

to

of

Well that looks OK

that

The thread about SCSI errors just stopped so I don't know what happened there.
There was some thought that maybe LVM was trying to access beyond the end of the
disk so you could check to see if the LV is using all the space on /dev/hdc and
back it off a little so see if that helps. It may also be worth upgrading to LVM 1.0.1rc4.

If it is the SCSI tape drive interfering then that's really odd because SCSI shouldnot interfere with IDE unles you have some interrupt clashes or perhaps
bad cabling causing signal noise.

Not very conclusive I'm afraid, but I did say I wasn't an IDE expert!

patrick

_______________________________________________
linux-lvm mailing list
[email protected]
http://lists.sistina.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html

What it's warning you about is reducing the size of the LV without first changing the size of the filesystem that is using it. eg (for reiserfs) you *must*

# resize_reiserfs <blah>
# lvreduce <blah>

Otherwise the filesystem will still think it is the same size as before and try to read/write off the end of the block device. This is almost guaranteed to cause problems.

You should reduce it by at least 1 Physical Extent (default is 4Meg), if you use the lower-case l to lvreduce it will do that for you (Capital L is for K/Meg/Gigabyes, lowercase l is for PEs) so

# lvreduce -l 1 /dev/LVM/mp3z

Will do the job *BUT YOU MUST RESIZE THE FILESYSTEM FIRST*

If you are using an ext2 filesystem then you can use the e2fsadm command:

# e2fsadm -l -1 /dev/LVM/mp3z

patrick

changing

aha... maybe I could encourage the man page's maintainer to make that point a little clearer?

I performed the above command expecting I would have to do an lvreduce later but it <gasp> did it for me... I haven't done a compare against backups to make sure I didn't lose data ok but I still have the problem... here's the output (but read beyond):

blocks

28.62 GB

now, in playing with this I've found that I get these error messages most consistently while shutting down. and analysis of /etc/init.d/halt shows that the errors are produced by an effort to remount / in readonly mode... the code reads like this:

by the time this code is run I've already un-mounted /var/LVM/mp3z and done a "vgchange -a n"... so in retrospect I can say that the only thing that has changed (I wasn't having these errors before installing LVM) is that I did a "pvcreate /dev/hdc"...

is the info above of any use and is there anything else I can try?

1k thx - e

-----Original Message-----
From: [email protected] [mailto:[email protected]] On
Behalf Of Patrick Caulfield
Sent: Friday, October 26, 2001 1:10 AM
To: [email protected]
Subject: Re: [linux-lvm] read_intr errors

when

What it's warning you about is reducing the size of the LV without first changing
the size of the filesystem that is using it. eg (for reiserfs) you *must*

# resize_reiserfs <blah>
# lvreduce <blah>

Otherwise the filesystem will still think it is the same size as before and try
to read/write off the end of the block device. This is almost guaranteed to cause problems.

me

You should reduce it by at least 1 Physical Extent (default is 4Meg), if you use the lower-case l to lvreduce it will do that for you (Capital L is for K/Meg/Gigabyes, lowercase l is for PEs) so

# lvreduce -l 1 /dev/LVM/mp3z

Will do the job *BUT YOU MUST RESIZE THE FILESYSTEM FIRST*

If you are using an ext2 filesystem then you can use the e2fsadm command:

# e2fsadm -l -1 /dev/LVM/mp3z

patrick

_______________________________________________
linux-lvm mailing list
[email protected]
http://lists.sistina.com/mailman/listinfo/linux-lvm
read the LVM HOW-TO at http://www.sistina.com/lvm/Pages/howto.html

I've added some extra words to the lvreduce man page to help clarify matters, and make it a little less scary!

It'll be fine. e2dsadm does the filesystem AND LV changes in the right order.

Don't reboot?

I'm out of ideas now, I'll have think....

patrick