Skip to contentSkip to conversation

[2.6.27.6] jfs on raid1 => attempt to access beyond end of device.

DISCUSSION Paweł Sikora 2 participants 2 messages Nov 20, 2008 1 message not in the archive

lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected]

#1 lore raw

20/11/2008, "Neil Brown" <[email protected]> napisał/a:

Quoted 35 lines

Hi.
I sent this reply, but I bounced. I don't know why but I'm sending it again a different way. Hopefully it will get through. Please Cc any followup to [email protected]. I think the email got through to the list.
NeilBrown

Quoted 6 lines · Paweł Sikora

hi,

few hours ago i've set up jfs filesystems on raid1 and raid0. during restoring backup i've got an errors in dmesg. the testcase on my system is easy and 100% reproducible:
just do the following command on jfs/raid1 device:

....

Quoted 1 line

/dev/sda2 * 748 2206 11719417+ fd Linux raid

                                          ^^^^^^^^^

size of sda2 is Kilobytes - 23438835 sectors.

Quoted 3 lines

md0 : active raid1 sda2[0] sdb2[1]

      11719296 blocks [2/2] [UU]
       ^^^^^^^^

size of md0 in kilobytes - 23438592 sectors.

Quoted 2 lines
[ 5583.796222] attempt to access beyond end of device
[ 5583.796227] md0: rw=1, want=23438832, limit=23438592

Largest 'want' value.

Quoted 8 lines

'want' is just less than size of sda2 'limit' is exactly size of md0 (no surprise there).

Quoted 1 line

any ideas what's wrong?

I suspect you created the filesystem on /dev/sda2, not realising that when you created a raid1 from sda2 and sdb2 it would be slightly smaller than sda2 (as md used up to 120K for metadata storage).

thanks for the quick reply!
afair i've ran mkfs.jfs on /dev/md/0. quick test...

working raid0 device:

# fsck.jfs -f -n /dev/md/1
fsck.jfs version 1.1.13, 17-Jul-2008
processing started: 11/20/2008 13.38.32
Filesystem is currently mounted.
WARNING: Checking a mounted filesystem does not produce dependable
results.
The current device is:  /dev/md/1
Block size in bytes:  4096
Filesystem size in blocks:  51757376
**Phase 0 - Replay Journal Log
**Phase 1 - Check Blocks, Files/Directories, and  Directory Entries
**Phase 2 - Count links
(...)

failing raid1 device:

# fsck.jfs -f -n -v /dev/md/0
fsck.jfs version 1.1.13, 17-Jul-2008
processing started: 11/20/2008 13.34.54

/dev/md/0 is mounted and the file system is not type JFS. (...)

and the raw sda2 device:

# fsck.jfs -f -n -v /dev/sda2
fsck.jfs version 1.1.13, 17-Jul-2008
processing started: 11/20/2008 13.34.50
The current device is:  /dev/sda2
Open(...READ/WRITE EXCLUSIVE...) returned rc = 0
Invalid magic number in the superblock (P).
Invalid magic number in the superblock (S).

The superblock does not describe a correct jfs file system. (...)

#2 lore raw
Quoted 9 lines · Pawe³ Sikora

20/11/2008, "Neil Brown" <[email protected]> napisa³/a:

failing raid1 device:

# fsck.jfs -f -n -v /dev/md/0
fsck.jfs version 1.1.13, 17-Jul-2008
processing started: 11/20/2008 13.34.54

/dev/md/0 is mounted and the file system is not type JFS.

Try
grep md /proc/mounts
and

    fsck.ext3 -f -n /dev/md/0

I looked again at the logs and there is evidence that it is an ext3 filesystem that was triggering those errors. Maybe you never did a mkfs after creating the raid1???

NeilBrown