It's postgres version 16. I have two questions on alerting as below.
1)If we want to have alerting on any node/instance that gets crashed :- In other databases like Oracle the catalog Views like "GV$Instance" used to give information on whether the instances are currently active/down or not. But in postgres it seems all the pg_* views are instance specific and are not showing information on the global/cluster level but are restricted to instance level only. So is there any other way to query the pg_* views to have alerts on the specific instance crash?
In Postgresql, cluster == instance. That's a historical fluke which might never go away. Thus, if the cluster is down, you can't access anything.
Connection poolers that use virtual IP addresses and are the modern definition of "cluster" sit on top of individual PG clusters. Even though the pooler auto-fails the (modern) cluster to the replica instance, PG still thinks one cluster is down, and the former-replica cluster is now the primary cluster.
Confusing? Yes. Just accept that PG cluster == instance, and that Postgresql is not Oracle.
2)Is there a way to fetch the data from pg_* view to highlight the specific connection/session/sqls which is using high memory in postgres?
