swap_pager: indefinite wait buffer

[Luciano Mannucci]

Hello everybody!

Since my last upgrade (to 11.3-RELASE-p6) I've noticed some messages
in the daily security report of one of my PPC 64 virtual servers, that
runs under KVM/Qemu, which says:

Feb 26 20:17:53 Saguaro kernel: swap_pager: indefinite wait buffer: bufobj: 0, blkno: 1698063, size: 20480

(numbers vary, of course)
It happens only under heavy IO.
I can reproduce this by launching a backup (I use bacula). I moved the
swap to cooked files, just in case it was a hardware problem, which
gave me only more messages of this kind. I know that the handbook
recommends to check hardware & cables, but being on a virtual machine
makes it a bit nonsensical. The machine does'nt lack ram, it has 8 CPUs
and 8 Gigabytes of ram. This is vmstat:

procs  memory       page                    disks     faults         cpu
r b w  avm   fre   flt  re  pi  po    fr   sr md99 md98   in    sy    cs us sy id
0 0 0  14G  1.2G   687   8   1   0   747  951   0   0    0  1622   704  3  1 95

and swapinfo says:

Device          1K-blocks     Used    Avail Capacity
/dev/md99         2097152   420564  1676588    20%
/dev/md98         2097152   419352  1677800    20%
Total             4194304   839916  3354388    20%


Is there something I should check?
Any kernel variables tunable via sysctl?

Thanks for any answer,

Luciano.
-- 
 /"\                         /Via A. Salaino, 7 - 20144 Milano (Italy)
 \ /  ASCII RIBBON CAMPAIGN / PHONE : +39 2 485781 FAX: +39 2 48578250
  X   AGAINST HTML MAIL    /  E-MAIL: posthamster at sublink.sublink.ORG
 / \  AND POSTINGS        /   WWW: http://www.lesassaie.IT/