Tobias Wolf
Tobias Wolf
Hi @MiningPickaxe You're right that `largest_conference` has a `jvb_instance` in its labels: ``` # curl -s 10.6.28.11:8999/metrics | grep largest # HELP largest_conference The current number of participants in the...
@MiningPickaxe since we are running this standalone I just configured a static config for Prometheus: ``` # cat /etc/prometheus/jobs/static/jitsi.yml - targets: - '10.6.28.11:8999' labels: job: "jitsi" instance: "h7020.example.com" ``` EDIT:...
Yes, a go-temptated string would probably the best. At first I thought instead of `k8s-aix02.kube-system.po-kube-apiserver-kvm335` it should just be `kube-apiserver-kvm335`, but maybe that is not a good idea. If, e.g.,...
I think so, yes. Something like `zone = "common"` https://www.icinga.com/docs/icinga2/latest/doc/12-icinga2-api/#api-objects-and-cluster-config-sync
Would it be possible to somehow ignore the old events and start the watch with new events only? I'm not interested in the past events, only new ones. There's [settings...
This is not bottlenecked by CPU, there are just way too many events in the namespace when it starts watching.
We made a special build which extends the timeout to 10 minutes, and then it works. But I would prefer if it would just ignore the old events.
I could use input from the maintainers on whether they would accept this simple fix and whether it needs to be confiurable.
The initial sync of the informer was definitely on the order of minutes. That's because in some namespaces we had hundreds of thousands of events. So to skip the backlog...