grafana issue postgresql_details.json

[Eric-zch]

It seems there is a small issue with postgresql_details.json.
dbname=~ should be used to replace datname=~.

Below is what I changed

"expr": "sum(irate(ccp_stat_database_xact_commit{pg_cluster=\"[[cluster]]\",pod=~\"[[pod]]\",dbname=~\"[[datname]]\"}[5m])) + sum(irate(ccp_stat_database_xact_rollback{pg_cluster=\"[[cluster]]\",pod=~\"[[pod]]\",dbname=~\"[[datname]]\"}[5m]))",
"expr": "sum(irate(ccp_pg_stat_statements_total_calls_count{pg_cluster=\"[[cluster]]\",pod=~\"[[pod]]\",dbname=~\"[[datname]]\"}[5m]))",
"expr": "sum(irate(ccp_stat_database_tup_fetched{pg_cluster=\"[[cluster]]\",pod=~\"[[pod]]\",dbname=~\"[[datname]]\"}[5m]))",
"expr": "sum(irate(ccp_stat_database_tup_inserted{pg_cluster=\"[[cluster]]\",pod=~\"[[pod]]\",dbname=~\"[[datname]]\"}[5m]))",
"expr": "sum(irate(ccp_stat_database_tup_updated{pg_cluster=\"[[cluster]]\",pod=~\"[[pod]]\",dbname=~\"[[datname]]\"}[5m]))",
"expr": "sum(irate(ccp_stat_database_tup_deleted{pg_cluster=\"[[cluster]]\",pod=~\"[[pod]]\",dbname=~\"[[datname]]\"}[5m]))",
"expr": "sum(irate(ccp_stat_database_tup_returned{pg_cluster=\"[[cluster]]\",pod=~\"[[pod]]\",dbname=~\"[[datname]]\"}[5m]))",
"expr": "sum(rate(ccp_stat_database_deadlocks{pg_cluster=\"[[cluster]]\",pod=~\"[[pod]]\",dbname=~\"[[datname]]\"}[5m]))",
"expr": "sum(rate(ccp_stat_database_conflicts{pg_cluster=\"[[cluster]]\",pod=~\"[[pod]]\",dbname=~\"[[datname]]\"}[5m]))",
"expr": "sum(irate(ccp_stat_database_xact_commit{pg_cluster=\"[[cluster]]\",pod=~\"[[pod]]\",dbname=~\"[[datname]]\"}[5m]))",
"expr": "sum(irate(ccp_stat_database_xact_rollback{pg_cluster=\"[[cluster]]\",pod=~\"[[pod]]\",dbname=~\"[[datname]]\"}[5m]))",
"expr": "sum by (mode) (ccp_locks_count{pg_cluster=\"[[cluster]]\",pod=~\"[[pod]]\",dbname=~\"[[datname]]\",mode=\"accessexclusivelock\"})",
"expr": "sum by (mode) (ccp_locks_count{pg_cluster=\"[[cluster]]\",pod=~\"[[pod]]\",dbname=~\"[[datname]]\",mode=\"exclusivelock\"})",
"expr": "sum by (mode) (ccp_locks_count{pg_cluster=\"[[cluster]]\",pod=~\"[[pod]]\",dbname=~\"[[datname]]\",mode=\"rowexclusivelock\"})",
"expr": "sum by (mode) (ccp_locks_count{pg_cluster=\"[[cluster]]\",pod=~\"[[pod]]\",dbname=~\"[[datname]]\",mode=\"sharerowexclusivelock\"})",
"expr": "sum by (mode) (ccp_locks_count{pg_cluster=\"[[cluster]]\",pod=~\"[[pod]]\",dbname=~\"[[datname]]\",mode=\"shareupdateexclusivelock\"})",
"expr": "sum by (mode) (ccp_locks_count{pg_cluster=\"[[cluster]]\",pod=~\"[[pod]]\",dbname=~\"[[datname]]\",mode=\"accesssharelock\"})",

[benjaminjb]

Oh, thanks @Eric-zch, I'm looking into that now and opened a PR for upstream pgmonitor (from where we pull these dashboards, etc.)

[benjaminjb]

Hey @Eric-zch, how are you installing the monitoring stack? I ask because I updated the dashboard in this repo (in the devel branch currently), so if you're installing directly from here (on that branch) using kustomize, everything should be good. I'm working to get that change merged into our main branch and also working to get the Helm chart updated in our OCI registry, but that might not be soon.

Also, just wanted to say thanks again for raising this issue.

[benjaminjb]

This should be fixed now, though don't be alarmed if you see datname in that dashboard still -- some of our metrics actually have datname as the label, while others have dbname. I've got a ticket to see if I can squash that issue, but for now at least, this dashboard works.