PG Monitoring review Dashboard 1

Related to the current monitoring we need to review:

• Dashboard #1 (discussed above):
  • ✅ TPS (transactions per second)
  • ⛔ QPS (queries per second)
  • ⛔ latency (average query duration)
  • ⛔ longest transaction time
  • ✅ Connections (total count and by state from pg_stat_activity)
  • ✅ Error count (ROLLBACKs absolute value)
  • ⛔ COMMIT vs. ROLLBACK ratio
  • ⛔ Buffer pool efficiency – hits vs. reads ratio
  • ⛔ Database IO time
  • ⛔ Temp bytes written
  • ✅ Replication lag
  • ⛔ Count of WALs waiting to be archived (a.k.a. archive_command lag) // there is "seconds since last WAL archive" in the old WAL-E dashboard, but its meaning is different that "lag size"
  • ⛔ WAL generation rates // there is WAL-G completed per second on the old WAL-E dashboard, but it's not generation - it's archiving
  • ✅ Locks (exclusive locks count)
  • ✅ Wraparound: Transaction ID age (exists in the "patroni: Overview" dashboard)
  • ⛔ Wraparound: Transaction Multixact ID age
  • ⛔ Database Size // Nik: I'm sure I saw it somewhere many times but cannot find now

Acceptance criteria:

• Fix the metrics and the dashboard