Make sure dispatch lag gets logged/graphed somewhere. Like graphite.wikimedia.org .
See Also:
T47533: [Story] Add monitoring interface to wikibase
Make sure dispatch lag gets logged/graphed somewhere. Like graphite.wikimedia.org .
See Also:
T47533: [Story] Add monitoring interface to wikibase
Status | Subtype | Assigned | Task | ||
---|---|---|---|---|---|
Invalid | None | T108944 [Epic] Improve change dispatching | |||
Invalid | None | T108931 [Epic] Improve metrics and statistics for wikidata | |||
Resolved | Addshore | T75647 [Story] Log/graph dispatch lag |
I used to graph this... years ago.. but I dont any more.
I would love to see this on graphite, ganglia, or anywhere
Currently being tracked on graphite using P2298 running on stat1002.
This could be run from anywhere on the cluster.
See basic graphs at:
https://graphite.wikimedia.org/render/?width=586&height=308&target=wikidata.dispatch.*.lag
https://graphite.wikimedia.org/render/?width=586&height=308&target=wikidata.dispatch.*.pending
We could possibly productionize this?
We should productionize this. Any idea how? Is there a trigger in Wikibase where we could compute these or something similar and send to statsd? Or should we put it as an interim solution into limn-wikidata-data?