I've been seeing subtle corruption on my RAID 5 drives using 2.2.17 kernel
with the 0.90 RAID patches (and ide patches as well). It's happened once
before and I had to reboot and fsck - it's starting to happen again.

I'm running Mandrake 7.1 on an ASUS A7V (onboard VIA Technologies VT 82C586
Apollo IDE (rev 16) and an onboard Promise PDC20265) with 3 x 30GB IBM
DTLA-307030 hard drives.

This machine was deployed in October/November and has been running without
incident since. It's mostly idle - it's a web/mail/dns/db/development server
for about 20 people - here's some MRTG graphs of how busy it actually is:

<http://staff.javelindigital.com/darron/images/uber-day.png>
<http://staff.javelindigital.com/darron/images/uber-week.png>
<http://staff.javelindigital.com/darron/images/uber-month.png>

I'm starting to see these in the logs:

May 31 04:01:13 uber kernel: EXT2-fs warning (device md(9,1)):
ext2_free_inode: bit already cleared for inode 280505
May 31 04:02:15 uber kernel: EXT2-fs warning (device md(9,1)):
ext2_free_inode: bit already cleared for inode 280505

How can I tell what drive/partition these warnings are referring to? Is it a
specific drive? A specific filesystem?

Find shows that that inode number is related to these files:

[root@uber /]# find . -inum 280505
./usr/lib/xemacs/xemacs-packages/lisp/gnus/nnspool.el
./usr/local/src/proftpd-1.2.1-0531

I think that's bad that those 2 files are sharing an inode but am not
entirely sure.

These messages only show up at 4 am when a couple disk intensive cronjobs
run but I have to take the whole machine down to runlevel 1 to fsck the
various mounted devices.

I'm going to upgrade to 2.2.19 with the mingo's RAID patches and the more
current ide patchset to see if that helps - does anyone else have any ideas?
-- 
Darron
[EMAIL PROTECTED]

-
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to [EMAIL PROTECTED]

Reply via email to