Hi, MoritzSorry for not reply last mail from Niv Sardi on 08 Aug 2008 in this thread, I must miss that mail.
On 2008-11-22, at 21:24, Moritz Muehlenhoff wrote:
On Mon, Jun 18, 2007 at 12:17:39AM +0800, Chun Tian (binghe) wrote:Package: linux-image-2.6.18-4-amd64 Version: 2.6.18.dfsg.1-12etch2 Severity: important Hi,Recently, I meet many times on many servers, large (>1TB) XFS filesystem throw kernel internal error:Filesystem "cciss/c0d2": XFS internal error xfs_trans_cancel at line 1138 of file fs/xfs/xfs_trans.c. Caller 0xffffffff881df006Does this error still occur with more recent kernel versions?
Now we're using linux-image-2.6.18-6-amd64 package from etch update. Seems this issue still happens sometimes but very rare than before. We have 200+ Debian box (etch) and 1000+ XFS filesystems installed, since not all nodes' kernel package are up to date, I can say the new etch 2.6.18 kernel fix this issue.
If you're running Etch, could you try to reproduce this bug with the 2.6.24 based kernel added in 4.0r4? http://packages.qa.debian.org/l/linux-2.6.24.html
Yes we're still running etch, and we may not upgrade to lenny soon even it's released.
OK, I'll try to upgrade, say, about 100 servers, before the end of this year, and see how many times the XFS internal error would happen in, say, one month, then report back.
We're running a big Web site in China and I personal am a little busy these days, so please give me more time since this bug report is already quite old:)
Cheers, Moritz
-- Chun Tian (binghe) NetEase.com, Inc. P. R. China