: Also got the same program with the swap space. However I found out that my
: Apache 1.0.0 daemons were spawning all over the G-D-PLACE. Console was
: unusable as the /var/log/messages were filling it up. So I broke the
Same problem. Linux 2.0.25 (much and safely matched). Scenario: something
sends apache wild (1.2.4) and it spawns 251 copies (yes, maxchild = 100
in the config!), fills up 128M of swap, and you get "accessing
nonexistent swap .." messages. I posted on linux-kernel to see if there
is an off-by-one error or similar in the swap code but nobody came up
with anything concrete.
The machine held up but dmesg was not pretty.
I have since reduced the amount of declared swap space to 127M within the 132M
partition in which it resides, just in case. And apache is being killed
and restarted every morning for good luck. No reoccurrence yet. But I have
seen this many times before and never resolved the cause. I feel it is to do
with stuck sockets and use of the -1 signal to apache. I rotate the logs in the morning at 8 and apache went crackers about 8.45 as far as I can tell. Maybe
a portscan? Apache should never have spawned that many children anyway.
This is on a scsi server with buslogic (not flashpoint) controller. Of course
it would happen when I was subbing for the normal solaris server for a while
(one broken cpu or scsi controoler) ...
: However I recall something in the newsgroups some time back and also in
: the SUSE 5.0 documentation about the buffer code in 2.0.29 and above being
: unreliable.
I am below that (and staying there!).
: Many thanx in advance.
De nada.
-- --------------------------------------------------------------------- Peter T. Breuer Phd. Ing., Area de Ingenieria Telematica E-mail: ptb@it.uc3m.es Dpto. Ingenieria Tel: +34 1 624 99 53 Universidad Carlos III de Madrid Fax: +34 1 624 94 30/65 Butarque 15, E-28911 Leganes URL: http://www.it.uc3m.es/~ptb Spain