[Tilesathome] Munin graphs for ROMA

[Milenko]

> -----Original Message-----
> From: tilesathome-bounces at openstreetmap.org [mailto:tilesathome-
> bounces at openstreetmap.org] On Behalf Of Mathieu Arnold
> Sent: Monday, December 08, 2008 3:33 PM
> To: tilesathome at openstreetmap.org
> Subject: Re: [Tilesathome] Munin graphs for ROMA
> 
> +--On 8 décembre 2008 12:25:43 -0500 Milenko <milenko at king-nerd.com>
> wrote:
> | Anyone have any ideas on how we could get some munin graphs showing
> some
> | statistics for the ROMA servers?
> |
> | Requests per hour - average request time - etc?
> |
> | Building the plugin looks fairly straightforward, but I'm not sure
> what
> | the easiest way to get the data out would be.
> 
> The lb stats may be of use :
> 
> http://api1.osm.absolight.net:8080/haproxy?stats
> and
> http://api1.osm.absolight.net/munin/osm.mat.cc/api1.osm.mat.cc.html#Hap
> roxy
> 
> --
> Mathieu Arnold

There's something interesting in those stats.  Note how the l0l0's server is set to 5 maxinstances, while the other are at 7.  Then check the total sessions and byte's in/out and note how they're higher for l0l0's server.  I've been wondering if there's a point where too many requests exponentially reduce the efficiency of the ROMA script due to overloading of the IO system.  So lowering maxinstances would not only decrease the response time of each individual request, but also increase the overall throughput of the server.

The stats on my server seem to suggest this - since you reduced my server to 7 from 9 the load average plummeted from ~8 to ~2 and outbound bandwidth doubled, which suggests that the server was handing out more requests than before.

-Jeremy