Now that fundraising is no longer using ganglia we can uninstall it from the fleet (gmetad / gmond / etc) and remove the relevant puppet bits.
- Mentioned In
- T183917: remove cloud VPS project 'ganglia'
T166322: spam from phabricator in labs
T183209: decom uranium
T175150: Decommission stat1003.eqiad.wmnet
T178220: Fix cronspam from /usr/local/sbin/pdns_gmetric
- Mentioned Here
- T169600: Enable diamond PowerDNSRecursor collector on dnsrecursors
T147426: Port gdnsd statistics from ganglia to prometheus
T145659: Port application-specific metrics from ganglia to prometheus
T177196: Port non-deprecated Diamond collectors to Prometheus
Alright, Ganglia is purged from everything across the board, except 17 hosts now! :) They are:
4 x maps codfw (osm/postgres)
4 x maps eqiad (osm/postgres)
3 x maps-test codfw (osm/postgres)
3 x labsdb eqiad (postgres)
2 x install (aggregators eqiad/codfw)
1 x ganglia-web (uranium)
All else DONE
Cool! Thanks for confirming. I had just kept it because there was no replacemente for postgres stats yet but that's also work in progress and coming up. Removed from maps and maps-test cluster just now :)
Ganglia has been uninstalled from the fleet, the aggregators are gone, the roles and the module is deleted, the DNS name is removed, for all purposes it's gone. The remaining hits for grepping "ganglia" across the repo are mostly related to the "ganglia_clusters" variable in Hiera which we should replace with LVS config or rename:
And then it appears a couple times in modules/confluent/manifests/kafka/mirror/jmxtrans.pp and in an example in wmflib.
But none of this means Ganglia is still running at any capacity, so this ticket is resolved.