[mythtv-users] XFS crash

Tim Tait tim.tait1 at gmail.com
Tue Feb 21 18:25:51 UTC 2006


Yan Seiner wrote:

>Per the docs, I set up my myth partition as xfs....  That lasted about a 
>week; tonight I got this:
>
>Feb 20 17:08:50 tooth kernel: XFS internal error XFS_WANT_CORRUPTED_GOTO 
>at line 1617 of file fs/xfs/xfs_alloc.c.
>Caller 0xf8926bda
>Feb 20 17:08:50 tooth kernel:  [<f8925a87>] 
>xfs_free_ag_extent+0x3f7/0x680 [xfs]
>Feb 20 17:08:50 tooth kernel:  [<f8926bda>] xfs_free_extent+0xda/0x100 [xfs]
>Feb 20 17:08:50 tooth kernel:  [<f8926bda>] xfs_free_extent+0xda/0x100 [xfs]
>Feb 20 17:08:50 tooth kernel:  [<f89804c9>] kmem_zone_alloc+0x49/0xb0 [xfs]
>Feb 20 17:08:50 tooth kernel:  [<f8980556>] kmem_zone_zalloc+0x26/0x50 [xfs]
>Feb 20 17:08:50 tooth kernel:  [<f8938547>] xfs_bmap_finish+0x107/0x190 
>[xfs]
>Feb 20 17:08:50 tooth kernel:  [<f895ea9f>] 
>xfs_itruncate_finish+0x20f/0x3c0 [xfs]
>Feb 20 17:08:50 tooth kernel:  [<f89756db>] xfs_trans_ijoin+0x2b/0x80 [xfs]
>Feb 20 17:08:50 tooth kernel:  [<f897b5ce>] xfs_inactive+0x45e/0x4b0 [xfs]
>Feb 20 17:08:50 tooth kernel:  [<f8989ac8>] linvfs_clear_inode+0xa8/0xc0 
>[xfs]
>Feb 20 17:08:50 tooth kernel:  [<c01640d5>] 
>invalidate_inode_buffers+0x15/0x90
>Feb 20 17:08:50 tooth kernel:  [<c017bf04>] clear_inode+0x104/0x150
>Feb 20 17:08:50 tooth rsyncd[30781]: rsync: mkstemp 
>"/home/samba/ACTD/Job/Johnson/.JOHNSON-8.0.DWG.xQtdsZ" (in tnd) failed: 
>Input/output error (5)
>Feb 20 17:08:50 tooth kernel:  [<c017ce49>] generic_delete_inode+0x129/0x130
>Feb 20 17:08:50 tooth kernel:  [<c017d013>] iput+0x53/0x70
>Feb 20 17:08:50 tooth kernel:  [<c0179e19>] dput+0xf9/0x1f0
>Feb 20 17:08:50 tooth kernel:  [<c0173b73>] sys_rename+0x1b3/0x1d0
>Feb 20 17:08:50 tooth kernel:  [<c0107960>] do_gettimeofday+0x20/0xc0
>Feb 20 17:08:50 tooth kernel:  [<c0102fdb>] sysenter_past_esp+0x54/0x75
>Feb 20 17:08:50 tooth kernel: xfs_force_shutdown(md8,0x8) called from 
>line 4125 of file fs/xfs/xfs_bmap.c.  Return
>address = 0xf898a6d9
>Feb 20 17:08:50 tooth kernel: Filesystem "md8": Corruption of in-memory 
>data detected.  Shutting down filesystem: md8
>Feb 20 17:08:50 tooth kernel: Please umount the filesystem, and rectify 
>the problem(s)
>
>This is not fatal but it's a real PITA; that partition has about 200 GB 
>of data.  I have it on an external USB drive but it takes hours to restore.
>
>How stable is XFS?  This crash occurred during an rsync.  I've used ext3 
>happily but changed per the docs....  Now I'm not so sure....
>
>This file system is on top of a RAID-5 array.  ISTR reading about 
>instability of XFS on top of raid...  Suggestions?  JFS?  ext3?  videotapes?
>
>--Yan
>  
>
Yan, you don't say what install you use, Redhat, Debian, etc or what kernel.

I have been happily using XFS on a 4 drive RAID5 config under Debian 
Testing with a 2.6.8 kernel for almost 2 years, w/o experiencing any XFS 
issues. I use some of it for /usr, some for /home with Samba mounts, and 
200GB for Mythtv.

Maybe you are having memory error problems?

Tim



More information about the mythtv-users mailing list