[DRBD-user] Bad entry in directory -->read-only
crosser at rol.ru
Wed Jan 18 09:33:24 CET 2006
There's been a long and elaborate discussion of this issue, and even
some investigations ;-)
Consensus seems to be that there is a hardware/timing issue on some(?)
chipsets, possibly related to DMA operation.
I could reproduce it on Supermicro SATA machines and no-name (Aquarius)
with Adaptec SCSI, but it never happend on Dells with Megaraid.
> I've gotten this problem about 4-5 times this past year where there is a
> bad directory entry in th ext3 file system that causes an aborted jornal
> causing a read-only file system.
> An hb_standby & hb_takeover fixes the problems. Whenever I do a chkdsk
> the filesystem has no errors, so it appears to be in memory (wasn't
> there another post on a similar issue?).
> Any inputs, suggestions, strategies to deal with the bug?
> Here are my log files:
> Jan 17 01:01:19 nfs1 kernel: EXT3-fs error (device drbd0): ext3_readdir:
> bad entry in directory #1574044: rec_len %% 4 != 0 -
> offset=0, inode=573445919, rec_len=12570, name_len=32
> Jan 17 01:01:19 nfs1 kernel: Aborting journal on device drbd0.
> Jan 17 01:01:19 nfs1 kernel: ext3_abort called.
> Jan 17 01:01:19 nfs1 kernel: EXT3-fs error (device drbd0):
> ext3_journal_start_sb: Detected aborted journal
> Jan 17 01:01:19 nfs1 kernel: Remounting filesystem read-only
> Jan 17 01:01:19 nfs1 kernel: journal commit I/O error
> Jan 17 01:03:23 nfs1 rpc.mountd: could not open /var/lib/nfs/rmtab for
> Jan 17 01:07:43 nfs1 rpc.mountd: could not open /var/lib/nfs/rmtab for
> Jan 17 01:12:34 nfs1 rpc.mountd: could not open /var/lib/nfs/rmtab for
> drbd-user mailing list
> drbd-user at lists.linbit.com
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 256 bytes
Desc: OpenPGP digital signature
Url : http://lists.linbit.com/pipermail/drbd-user/attachments/20060118/fd1a8d47/attachment.pgp
More information about the drbd-user