[OSM-talk] Linuxtag demonstration down - Re: handover of tile.openstreetmap.org

Artem Pavlenko artem at mapnik.org
Wed May 30 15:30:32 BST 2007


On 30 May 2007, at 15:22, Jon Stockill wrote:

> Nick Hill wrote:
>> Hi Jon
>>
>> I looked at apache2.conf.
>>
>> # prefork MPM
>> # StartServers ......... number of server processes to start
>> # MinSpareServers ...... minimum number of server processes which  
>> are kept spare
>> # MaxSpareServers ...... maximum number of server processes which  
>> are kept spare
>> # MaxClients ........... maximum number of server processes  
>> allowed to start
>> # MaxRequestsPerChild .. maximum number of requests a server  
>> process serves
>> <IfModule prefork.c>
>> StartServers        10
>> MinSpareServers     10
>> MaxSpareServers     20
>> MaxClients          60
>> MaxRequestsPerChild  100
>> </IfModule>
>>
>>
>> I reckon changing maxclients to 20 may help.
>>
>> With the settings above, 60 apache processes could start. We could  
>> have 40
>> processes being launched then being immediately killed as the load  
>> shoots up and
>> down. This can be very resource intensive. Apache tends to queue  
>> http requests
>> anyway until an http process can handle them.
>
> The easiest way to tune that is to look at how much ram everything but
> apache is using - work out how much you have to play with, then divide
> this available ram figure by the average size of your apache process -
> the number you arrive at is the number of max clients you should get
> away with before driving the machine into swap - at which point
> performance goes down the pan.
>
> That gives you a good starting point - obviously if there are apache
> modules you're not using they should be disabled to reduce the memory
> footprint and increase the number of instances you can run. If the  
> tasks
> are particularly cpu intensive you may find that you need to reduce  
> the
> number of processes anyway as you'll hit CPU limits before you run out
> of ram.
>
> -- 
> Jon Stockill
> lists at stockill.net
>

Starting a new thread per request rather then a process might speed  
things up a bit ( a lot), IMHO.
Assuming we're running apache2, how about 'worker' MPM ?

Cheers,

> _______________________________________________
> talk mailing list
> talk at openstreetmap.org
> http://lists.openstreetmap.org/cgi-bin/mailman/listinfo/talk
>

Artem Pavlenko
http://mapnik.org



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20070530/b7182fa3/attachment.html>


More information about the talk mailing list