Switched up to 2.6.6 and am noticing some problems in xfs:
1) the
normal xfs maintenance processes hung this morning: xfsdump hung during
the third fs (home) it was dumping, about the time that xfs_fsr kicked in
(at 3:15). Xfsdump started at 2:34am and bzip2 shows a cputime of
54:53, xfsdump: 34 seconds and xfs_fsr (2 copies) show 0 seconds
accumulated cputime. xfsdump is waiting on lock_p, one xfs_fsr is
also waiting on lock_p and the other is waiting on "wait4".
I'm assumming the 2 waiting on "lock_p" are the one's showing up
up in "top" as the load factor of 2 and 99% CPU spent in wait?
Regular file access to the /home partition seems unaffected.
2) Seems to have cropped up after install of 2.6.6 -- odd WARNings
in my log file:
Jun 9 02:34:17 ishtar kernel: pagebuf_get: failed to lookup pages
Jun 9 03:15:13 ishtar kernel: pagebuf_get: failed to lookup pages
Jun 9 03:15:13 ishtar last message repeated 3 times
Jun 9 03:30:44 ishtar kernel: pagebuf_get: failed to lookup pages
Jun 9 03:30:44 ishtar kernel: pagebuf_get: failed to lookup pages
...backwards
Jun 8 02:31:11 ishtar kernel: pagebuf_get: failed to lookup pages
Jun 8 02:31:11 ishtar last message repeated 2 times
...backwards
Jun 4 02:31:43 ishtar kernel: pagebuf_get: failed to lookup pages
Jun 4 02:31:43 ishtar last message repeated 18 times
Jun 5 02:32:08 ishtar kernel: pagebuf_get: failed to lookup pages
Jun 5 02:32:08 ishtar last message repeated 15 times
...backwards
Jun 3 04:05:14 ishtar kernel: pagebuf_get: failed to lookup pages
---
I installed 2.6.6 on Jun 1, I note that backups and fsr ran with no
warnings in the log, but every morning after that I seemed to have
gotten the error messages.
The warning messages didn't seem to be a problem, but the hanging of
the backup processes is definitely a "minus" in function...
Suggestions?
Revert to 2.6.5?
-l
|