Good morning everyone,
I’m starting to use PPM to monitor some databases we have on the AWS cloud,
in practice we have an Aurora PostgreSql Vers. 15.2 instance, with 15 databases inside.
We have installed and configured PPM Server 2.37.1
I’m trying to set up the databases so I can monitor them, so I followed the directions in the documentation,
but i’m probably doing something wrong.
When I add the instance, via “Add Service” → PostgreSql, I insert
Hostname XXXXXXXXX
Service name XXXXXXXX
Port 5432
Username AAAAAAA
Password BBBBBBB
Database (I enter the name of 1 of the 15 databases)
Max query length (NULL)
environment (WORK)
Region (NULL)
Availability Zones (NULL)
Replication set (NULL)
Clusters (NULL)
Custom labels (NULL)
Additional options (NONE)
For monitoring we have created a dedicated user (without associating the rds_superuser role)
CREATE ROLE percona_monitoring WITH
LOGIN
NOSUPERUSER
INHERIT
NO CREATEDB
NO CREATEROLE
NO REPLICATION
CONNECTION LIMIT 10
PASSWORD ‘BBBBBBBBBB’;
GRANT pg_monitor TO percona_monitoring;
The pg_state_statement is active on the instances, for monitoring
After confirming the entry, the service was not UP , but N/A due to the maximum number of connections that I brought to 30 at that point
and the service became green with status UP.
I immediately noticed on the DB side that the monitoring user used all the available connections, and trying to access QAN, going to the detail tables,
the message “too many connections” is displayed, connections which, looking at the db, result in all 30 being idle.
I realized that by adding this first service, the connections hooked up to all the db’s present, in some cases with 2/3 sessions per db.
So trying to add a new db of the other 14 left, it immediately tells me “too many connection”, and if I increase the maximum number of sessions to the percona_monitoring user
they are immediately occupied and it won’t let me add this DB.
I also set Metrics resolution, sec to RARE
I think there may be something unclear on my side in what I do, and I try to express my doubts:
Is it correct to have a dedicated user without rds_superuser role?
Is it correct to configure every single DB, by adding a new service and/or should I use other connection options for Aurora Postgres clusters?
I apologize if I put the post in the wrong area, but i’m new and it is my firs post
Thanks in advance to anyone who can help me clarify my ideas and solve the problem.
Have a nice Day
Ivan