Custom label for the etcd client service created by the etcd-operator, for Prometheus to scrape metrics from the etcd cluster pods.

5 views
Skip to first unread message

SANAM REDDY

unread,
Oct 30, 2019, 10:14:17 AM10/30/19
to CoreOS Dev
Assuming people are already using Prometheus with etcd-operator and etcd-cluster, is there any solution to define ServiceMonitor so that Prometheus can only select "etcd-cluster-client" service to retrieve stats from the etcd cluster pods??

I tried passing the default label "app=etcd" to Prometheus and this is working fine. But the label is attached to both client and peer services( which have common endpoints i.e. etcd cluster pods), Prometheus is scrapping metrics 2 times from each of the cluster pod. 

Following is the link for github issue:

Thanks,
Sarath.
Reply all
Reply to author
Forward
0 new messages