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

[Jon Burgess]

Feel free to change those numbers. You probably have more experience
at optimising web server settings than I do.

I know that we do get many short spikes of active processes when we
are doing the on-line tile rendering. I think we often get 30+
processes getting launched as we spend a couple of seconds rendering
new tiles. Just leave "vmstat 5" or similar running for a while on
tile and you should see what I mean. I'm not sure if this is lots of
processes all wanting to do rendering or just some single threading in
mod_ruby causing requests to back up.

It would be good if we could do something to smooth this out although
I think things have been running OK the past week that the online
rendering has been enabled. The munin cpu stats don't look much
different over the past week compared to previous.

One easy knob to tweak is the max active requests before disabling
online rendering and just returning a 404 tile in cat_tile.rb (just a
hard coded number, currently 64 if I remember correctly).

My previous experience was that if you edit
/home/www/tile/ruby/cat_tile.rb then you should restart the web server
otherwise I think mod_ruby can get confused.

  Jon


On 30/05/07, Nick Hill <nick at nickhill.co.uk> 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.
>
>
> Jon Burgess wrote:
> > Over the past week or so i've seen issues with loading the main
> > www.openstreetmap.org map page and no problems with the tiles
> > themselves. On several occasions i've waited 30+ seconds for the page
> > to appear.
> >
> > What failure did you see Joerg, the tiles or the page itself?
> >
> > In cases like this the information freeway site can be a good one to
> > try since it fetches the tiles without relying on
> > www.openstreetmap.org.
> >
> >  Jon
> >
>


-- 
    Jon