We only collect free / max / total memory on Blazegraph. It might make sense to also collect the GC time / count and the usual heap / non heap memory metrics (see elasticsearch for an example). Activating GC logs would help as well.
Description
Description
Details
Details
Subject | Repo | Branch | Lines +/- | |
---|---|---|---|---|
wdqs: cleanup old GC logs | operations/puppet | production | +10 -0 | |
wdqs: activate GC logs | wikidata/query/rdf | master | +23 -2 |
Related Objects
Related Objects
Event Timeline
Comment Actions
Elasticsearch configuration is done through puppet (https://github.com/wikimedia/puppet/blob/production/modules/elasticsearch/manifests/init.pp#L132-L141)
Comment Actions
Mentioned in SAL (#wikimedia-operations) [2017-03-06T19:52:42Z] <gehel> restarting wdqs-updater on wdqs* servers to activate GC logs - T159248