[users@httpd] Each httpd process consumes so much swap server crashes

[Stargazer <st...@blueyonder.co.uk>]

Hi

There are about 80 PHP virtual hosts, all using MySql, a mix of joomla, 
mantis, pligg etc. Its RH linux/Apache 2.2.8, Mem 1G, Swap 4G.
The server gets ~ 1 hit/sec across all hosts. About an hour after a 
restart top shows some swap being used which creeps up until free swap 
hits zero. I've got a 1 min cron job testing for this and restarting 
apache when it happens, so at least the server doesn't crash now but 
performance up to that point is terrible. After this restart performance 
rocks, but only while swap use is 0.

I've made various attempts at tuning httpd.conf. The only thing which 
seems to stop the swap problem is if MaxRequesrtsPerChild is 64! I'm no 
expert at this but realise that is too extreme: each site takes several 
seconds to load at that rate. Even at 100, the swap problem kicks in but 
takes longer.

What I'm really looking for is the reason top shows each http child to 
consume ~200m VIRT, ~80m RES (thats with the default httpd.conf 
settings, not my tuning). /apache-status shows about 12 child processes, 
always with SS values < 10. Are there any debug tools to let me see 
exactly what each httpd child is doing, ideally which site, db 
connections etc? Also, is my understanding correct here - these numbers 
*are* bad?
Or how about the bigger picture - am I asking too much of the server: 
dual core AMD Operon 1210, 2x 1.8Ghz ?

Thanks for any help

-- Eric
 


---------------------------------------------------------------------
The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:http://httpd.apache.org/userslist.html> for more info.
To unsubscribe, e-mail: users-unsubscribe@httpd.apache.org
   "   from the digest: users-digest-unsubscribe@httpd.apache.org
For additional commands, e-mail: users-help@httpd.apache.org