> Just a little update, this is what's going on over the server:
>
> 11:59am up 16:12, 1 user, load average: 10,51, 5,30, 2,47
> 151 processes: 144 sleeping, 6 running, 1 zombie, 0 stopped
> CPU states: 6,5% user, 2,1% system, 0,0% nice, 91,3% idle
> Mem: 449484K av, 444476K used, 5008K free, 0K shrd, 12K
buff
> Swap: 265032K av, 263440K used, 1592K free 24896K
cached
>
> PID USER PRI NI SIZE RSS SHARE STAT %CPU %MEM TIME COMMAND
> 9295 spamd 15 0 90604 79M 4788 S 0,0 18,1 0:09 spamd
> 9293 spamd 15 0 90228 78M 4908 S 0,0 17,9 0:09 spamd
> 9289 spamd 18 0 90248 66M 3924 S 0,0 15,2 0:09 spamd
> 9298 spamd 15 0 22808 14M 4764 S 0,0 3,2 0:00 spamd
> 9296 spamd 15 0 22904 14M 4744 S 0,0 3,2 0:00 spamd
> 9291 spamd 15 0 22896 14M 4732 S 0,0 3,2 0:00 spamd
> 9297 spamd 15 0 23008 14M 4740 S 0,0 3,2 0:00 spamd
> 9292 spamd 16 0 22480 13M 4856 S 0,0 3,1 0:00 spamd
> 9294 spamd 15 0 22460 13M 4800 S 0,0 3,0 0:00 spamd
> 9299 spamd 15 0 22440 13M 4908 S 0,0 3,0 0:00 spamd
> 9302 spamd 15 0 22808 12M 4636 D 0,0 2,9 0:00 spamd
> 9300 spamd 15 0 22524 12M 4720 D 0,7 2,8 0:00 spamd
> 9301 spamd 15 0 22460 12M 4652 D 0,1 2,7 0:00 spamd
> 9290 spamd 15 0 22704 9628 3760 S 0,0 2,1 0:00 spamd
> 7844 apache 15 0 7708 7204 6336 S 0,0 1,6 0:04 httpd
> 7437 apache 15 0 7488 7008 4916 S 0,1 1,5 0:04 httpd
> 7564 apache 15 0 7520 7008 4840 S 0,0 1,5 0:04 httpd
>
> I mean, 80 MB for the master spamd porcess? How do I limit this?
Well, the oldest (I don't know about 'master') spamd there is 9289, which is
only using 66M. And to all intents, you only have 3 spamd processes doing
anything, the rest are just sitting there. I'm not sure what is creating 14
of the beasts, but it looks like you could probably get away with 6 or so
just fine at that workload. That would save you 80M or so.
80M doesn't strike me as unusual for spamd if you have any of the addon
rulesets. So I'd suggest you don't limit the 80M if you want to catch spam.
Instead you work with it and perhaps limit the number of 80M instances.
One thing I notice is that you seem to be near the swap limit as well as
memory limit. I think you would be a lot better off if you weren't
completely out of swap space. Try taking swap up to some larger size, like
maybe at least the size of memory, rather than half the size of memory.
Loren
