After reading the list a bit, looked into the kernel config, and
confirmed it was using 8K stacks. I saw no errors on the underlying
ide devices in dmesg, but that doesn't really preclude the possibility
of a disk hardware error I suppose. I am currently running memtest86
on the machine to rule that out.
All in all it was frustrating, lucky it happened before the machine
went into production though. I was wanting to use XFS for a
mailserver with (lots of) messages stored in Maildir format; I'll
probably go back to ext2/mbox, it's what I'm familiar and comfortable
with. Given the kernel messages below, it appears the problem is with
XFS.
The machine had an XFS root filesystem, and the filesystem corruption
was such that it wouldn't even boot through to single user mode.
After booting from CD and running xfs_repair, the system works again,
but I'm reluctant to try again without really knowing what went wrong.
Alex
---- Previous message ----
My humble apologies for spamming the list, but my previous email was
stupidly not sent from the right machine, so I'm sending it again so
you'll have the correct reply address.
It would be fair to say also that the raid device went into degraded mode (I'm
not 100% sure whether before or after), and the box is really sick now. I'll
do some postmortem work later (I need sleep now) and post here if there is
interest.
Hi,
I had a problem with an XFS filesystem while creating a large Maildir
directory (converting mbox to Maildir). Can anyone tell me what the
following means (from kern.log)?
Mar 14 21:54:06 mail kernel: 0x0: 58 41 47 46 00 00 00 01 00 00 00 00 00 01 72 9
c
Mar 14 21:54:06 mail kernel: Filesystem "md0": XFS internal error xfs_alloc_read
_agf at line 2195 of file fs/xfs/xfs_alloc.c.
Caller 0xc41598ce
Mar 14 21:54:06 mail kernel: [__crc_elevator_init+3540111/6084970] xfs_alloc_re
ad_agf+0x14f/0x1dc [xfs]
Mar 14 21:54:06 mail kernel: [__crc_elevator_init+3538515/6084970] xfs_alloc_fi
x_freelist+0x112/0x38d [xfs]
Mar 14 21:54:06 mail last message repeated 2 times
Mar 14 21:54:06 mail kernel: [__crc_elevator_init+3540527/6084970] xfs_alloc_ve
xtent+0x113/0x39e [xfs]
Mar 14 21:54:06 mail kernel: [__crc_elevator_init+3713853/6084970] xfs_ialloc_a
g_alloc+0x1cb/0x63f [xfs]
Mar 14 21:54:06 mail kernel: [__crc_elevator_init+3863067/6084970] pagebuf_get+
0x81/0x13b [xfs]
Mar 14 21:54:06 mail kernel: [__crc_elevator_init+3819076/6084970] xfs_trans_re
ad_buf+0x285/0x2c6 [xfs]
Mar 14 21:54:06 mail kernel: [__crc_elevator_init+3720135/6084970] xfs_ialloc_r
ead_agi+0x94/0x12a [xfs]
Mar 14 21:54:06 mail kernel: [__crc_elevator_init+3715912/6084970] xfs_dialloc+
0x12a/0x82e [xfs]
Mar 14 21:54:06 mail kernel: [__crc_elevator_init+3602623/6084970] xfs_bmap_sea
rch_extents+0x45/0x4a [xfs]
Mar 14 21:54:06 mail kernel: [__crc_elevator_init+3772758/6084970] xlog_grant_l
og_space+0xfe/0x31b [xfs]
Mar 14 21:54:06 mail kernel: [__crc_elevator_init+3740199/6084970] xfs_ialloc+0
x4d/0x3dc [xfs]
Mar 14 21:54:06 mail kernel: [__crc_elevator_init+3768522/6084970] xlog_grant_p
ush_ail+0x3d/0x160 [xfs]
Mar 14 21:54:06 mail kernel: [__crc_elevator_init+3822396/6084970] xfs_dir_iall
oc+0x64/0x21c [xfs]
Mar 14 21:54:06 mail kernel: [__crc_elevator_init+3839899/6084970] xfs_create+0
x2f8/0x559 [xfs]
Mar 14 21:54:06 mail kernel: [__crc_elevator_init+3876884/6084970] linvfs_mknod
+0x1ab/0x322 [xfs]
Mar 14 21:54:06 mail kernel: [__crc_elevator_init+3688658/6084970] xfs_dir2_nod
e_lookup+0x93/0x9e [xfs]
Mar 14 21:54:06 mail kernel: [__crc_elevator_init+3659900/6084970] xfs_dir2_loo
kup+0xe2/0x10e [xfs]
Mar 14 21:54:06 mail kernel: [__crc_elevator_init+3817203/6084970] xfs_trans_un
locked_item+0x22/0x37 [xfs]
Mar 14 21:54:06 mail kernel: [__crc_elevator_init+3878316/6084970] linvfs_permi
ssion+0x1b/0x21 [xfs]
Mar 14 21:54:06 mail kernel: [__crc_elevator_init+3877278/6084970] linvfs_creat
e+0x13/0x17 [xfs]
Mar 14 21:54:06 mail kernel: [vfs_create+187/239] vfs_create+0xbb/0xef
Mar 14 21:54:06 mail kernel: [open_namei+328/1301] open_namei+0x148/0x515
Mar 14 21:54:06 mail kernel: [filp_open+44/73] filp_open+0x2c/0x49
Mar 14 21:54:06 mail kernel: [get_unused_fd+40/183] get_unused_fd+0x28/0xb7
Mar 14 21:54:06 mail kernel: [sys_open+49/103] sys_open+0x31/0x67
Mar 14 21:54:06 mail kernel: [syscall_call+7/11] syscall_call+0x7/0xb
Other info:
Linux version 2.6.8-1-386 (dilinger@xxxxxxxxxxxxxxxxxxxx) (gcc version
3.3.5 (Debian 1:3.3.5-2)) #1 Thu Nov 11 12:18:43 EST 2004, which is a
standard Debian (testing) kernel.
XFS filesystem was set up by the debian installer, no doubt with
default options. The filesystem in on a software raid device
(mirrored) The directory had ~5532 entries; the userspace program
died, and my first guess is that it may have hit some directory entry
limit?
# xfs_info /
meta-data=/ isize=256 agcount=8, agsize=94876
blks
= sectsz=512
data = bsize=4096 blocks=759008,
imaxpct=25
= sunit=0 swidth=0 blks,
unwritten=1
naming =version 2 bsize=4096
log =internal bsize=4096 blocks=2560, version=1
= sectsz=512 sunit=0 blks
realtime =none extsz=65536 blocks=0, rtextents=0
I'd be glad to supply any other information that may be helpful.
Cheers,
Alex
|