[collectd] apache and collectd plugin
bonbons at linux-vserver.org
Mon Mar 16 10:39:39 CET 2009
On Mon, 16 March 2009 Amit Gupta <amit.gupta221 at gmail.com> wrote:
> Thanks for your quick reply. One easy way to handle this is to have a
> separate lighttpd plugin which will be similar to apache's plugin and
> will handle lighttpd specific statistics, for instance, fastcgi
> statistics which apache mod_status doesn't expose.
> I will be happy to work on a separate lighttpd plugin if the separate
> plugin for lighttpd gets a green signal.
> The other approach would be to rename apache plugin to httpd plugin
> (as someone has already suggested) and have two sub blocks as apache
> and lighttpd (each of them will have have their separate
> configuration settings e.g URL, CAcert and so on).
> Which, according to you, is a better approach?
I would say neither is the best approach...
The best way to solve it would probably be to have apache plugin use
different configuration layout:
This way the <Host> blocks could be specified to denote different
servers (the code parsing status page would make distinction between
apache and lighttpd and collectd available statistics (as it does
already , at least partially)
The layout could also be like this (better?):
This would allow specifying the host name for which the apache plugin
collects statistics and differentiate between to web-server processes
on the same host with different instances.
Default instance would be "" and default host would be collectd's host
The configuration outside of <Server> block would continue working as
it does not but be handled as if it was enclosed with <Server> block
(and using default values for Host and Instance)
This kind of thought could also be applied to other plugins that can
collect remote data...
More information about the collectd