This message was deleted.
# rke2a
adamant-kite-43734
02/05/2024, 8:40 PMThis message was deleted.
c
creamy-pencil-82913
02/05/2024, 8:43 PMI suspect that’s not accurate. Are you sure you’re actually scraping etcd metrics? if etcd didn’t have a leader your cluster would be unusuable.
creamy-pencil-82913
02/05/2024, 8:44 PMthat other event is unrelated to etcd, I don’t know what the “operator-lock” lease is for but its not part of RKE2.
creamy-pencil-82913
02/05/2024, 8:44 PMwhat is the actual problem that you are troubleshooting
r
rapid-noon-15872
02/05/2024, 8:48 PMThis is a pretty fresh build, just has longhorn and added the monitoring app last week. Today I logged in and saw this cluster stuck on updating with a failure. I poked around and saw the previous screenshot, and then went into the cluster config, and checked the box to expose etct metrics, thinking maybe that was the only issue. after that change was applied, the clusters page shows this cluster as active. the previous screenshot etcd status remains.
rapid-noon-15872
02/05/2024, 8:49 PMI don't know if that was an auto update or what, I didn't trigger it today or anything.
rapid-noon-15872
02/05/2024, 8:52 PMI appear to have graphed metrics for all the default datapoints for etcd. I just do not know how to verify the election and fix the polling if it is incorrect
c
creamy-pencil-82913
02/05/2024, 8:53 PMyeah I don’t know where that dashboard is pulling that data from, but I can guarantee that if etcd didn’t have a leader, nothing on that cluster would be functional.
r
rapid-noon-15872
02/05/2024, 9:03 PMRoger that, thanks. It showed up after adding the Monitoring app from the Apps page. Probably a default Prometheus metric. I'll look at the Prometheus targets I guess.
c
creamy-pencil-82913
02/05/2024, 10:29 PMyou might need to add etcd-expose-metrics: true to the rke2 config to make them scrapable?