Top shows lots of qmail-local processes as waiting or sleeping, but
they pop off and are replaced pretty quickly by others (this is a mail
server). Tasks stays around 170, 99% of which are sleeping.
Additionally, the server is responsive... The previously-mentioned
associate and I even began to wonder if perhaps 'top' didn't count
differently the same load :) ?
Here is the output of vmstat (split across two lines for clarity)
procs -----------memory---------- ---swap--
r b swpd free buff cache si so
1 9 0 954120 0 1024624 0 0
-----io---- --system-- ----cpu----
bi bo in cs us sy id wa
32 25 9 31 9 7 63 20
Idle stays > 60% most of the time and Wait hangs around 20% to 30%.
It should be noted that this server is handling no more mail than it
did running 2.4.19. ACPI and APMS are not enabled. Only loaded
modules are 'sg' 'rtc' '8250' and 'serial_core'.
Henrique de Moraes Holschuh wrote:
On Mon, 28 Mar 2005, Aaron Thoreson wrote:
I have a few machines running 2.4.19. I've upgraded them to 2.6.9.
I run xfs and use the backports.org version of mod-init-tools,
procps and xfsprogs. I've had no problems to speak of, except on
some servers, where the load has gone from a low average 2 or 3 to
stabilizing at 12-14. An associate with a similar setup on his
machines saw the same thing.
What does atop (if you have accounting set up) or top says? What
about
vmstat ?
--
Aaron Thoreson
Network Group
Midcontinent Communications
aaront@midco.net
--
To UNSUBSCRIBE, email to debian-isp-REQUEST@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact
listmaster@lists.debian.org