from bug 51497c#1 (Antoine):
That would be very nice to have, specially the profiling bits. That project
does not have any documentation beside the puppet manifests though. Probably
lower priority compared to Icinga.
Version: unspecified
Severity: enhancement
The deployment-graphite instance is intended to do this monitoring. I started to work on it but haven't finished making the puppet modules/roles usable in beta. Graphite is running there along with txstatsd. Gdash is not present and configuration has not been changed in operations/mediawiki-config to point to the new server.
Graphite stats from deployment-prep are now sent to graphite.wmflabs.org. Labs Graphite is soon going to get 'real hardware' and then the data will just move there. The current setup is fully puppetized, pulling in stats via diamond.
I'll leave it to the deployment-prep folks on what to do with deployment-graphite :)
Reopening. Yuvi made diamond send host metrics to a central graphite.wmflabs.org instance. Now we need a dashboard on top of it to replace Ganglia and have some alerting system set up.
So... we have:
Marking as fixed.
+1 for the monitoring part of it. I am wondering what to do for MediaWiki profiling though. Should we use the same graphite instance or a dedicated one? I don't mind filing another task.