ClusterLabs / ha_cluster_exporter

Prometheus exporter for Pacemaker based Linux HA clusters
Apache License 2.0
79 stars 35 forks source link

No instance id inside export metrics #240

Closed sadon closed 6 months ago

sadon commented 10 months ago

Almost all of metrics have no 'instance' inside exported metrics as result default dashboards stay unusable.

image

example Online nodes query: count(ha_cluster_pacemaker_nodes{instance="", status="online"} == 1) / count(count(ha_cluster_pacemaker_nodes{instance=""}) by (node))

Exporter:

curl localhost:9664/metrics -s | grep "ha_cluster_pacemaker_nodes"
# HELP ha_cluster_pacemaker_nodes The status of each node in the cluster; 1 means the node is in that status, 0 otherwise
# TYPE ha_cluster_pacemaker_nodes gauge
ha_cluster_pacemaker_nodes{node="dc-test-pg-01",status="dc",type="member"} 1
ha_cluster_pacemaker_nodes{node="dc-test-pg-01",status="expected_up",type="member"} 1
ha_cluster_pacemaker_nodes{node="dc-test-pg-01",status="maintenance",type="member"} 0
ha_cluster_pacemaker_nodes{node="dc-test-pg-01",status="online",type="member"} 1
stefanotorresi commented 6 months ago

There has never been any "instance" label in any of the metrics. Where does this expectation come from?