Jeremy Jackson wrote:
Hi,
If anyone is interested furthur details can be forwarded.
This happened on my mail server after 30 days uptime. XFS is
SGI XFS snapshot-2.4.23-2003-12-01_00:33_UTC with ACLs, debug enabled
SGI XFS Quota Management subsystem
procmail processes started to accumulate stuck in lock_wait, had to reboot.
Mar 4 17:13:41 dmz kernel: __alloc_pages: 4-order allocation failed
(gfp=0xf0/0)
Mar 4 17:13:45 dmz last message repeated 6 times
.......
Mar 4 18:32:25 dmz kernel: XFS assertion failed:
xfs_bmbt_disk_get_startoff(r1) + xfs_bmbt_disk_get_blockcount(r1) <=
xfs_bmbt_disk_get_startoff(r2), file: xfs_btree.c, line: 287
partly decoded oops:
Just as an aside, running with debug enabled means that you are absolutely
crippling xfs performance, debug is really a developer tool, not a production
use flag.
It did however point to exactly where in the code it went belly up, in a
function which only exists when debug is turned on to check the
layout of some data structures. I would run xfs_check on the filesystem
and then xfs_repair, send the output of both to the list.
Steve
|