Postgres Backend Connections is WARNING

[TBT]

We've been seeing more of these lately: "POSTGRES_BACKENDS WARNING: (host:127.0.0.1) 76 of 100 connections (76%)".

What typically is the cause? What more can I do to investigate?

[rkennedy]

How many hosts / services do you have running in XI? This could just be growing as you add additional services with time.

[lmiltchev]

Where do you see this? In the logs? Is this a check that you are running? Can you show us the config?

I would guess that you are still using the "default" setting of 100 max connections in the postgresql.conf:

Code: Select all

max_connections = 100

You could bump up this value a bit, and restart postgres so that changes can take effect.

Code: Select all

service postgresql restart

[TBT]

How many hosts / services do you have running in XI? This could just be growing as you add additional services with time.

As mentioned, this is occurring on several of our XI servers. One for example has 146 Hosts and 1916 Services. Certainly nothing near the limitations of XI.

[TBT]

Where do you see this? In the logs? Is this a check that you are running? Can you show us the config?

I would guess that you are still using the "default" setting of 100 max connections in the postgresql.conf:

Code: Select all

max_connections = 100

You could bump up this value a bit, and restart postgres so that changes can take effect.

Code: Select all

service postgresql restart

- We receive the Service Alerts via email.
- We put these DB checks in place, with a 75% warning and 90% critical alarm.
- The /var/lib/pgsql/data/postgresql.conf has not been modified and currently contains "max_connections = 100"

[rkennedy]

Where do you see this? In the logs? Is this a check that you are running? Can you show us the config?

I would guess that you are still using the "default" setting of 100 max connections in the postgresql.conf:

Code: Select all

max_connections = 100

You could bump up this value a bit, and restart postgres so that changes can take effect.

Code: Select all

service postgresql restart

- We receive the Service Alerts via email.
- We put these DB checks in place, with a 75% warning and 90% critical alarm.
- The /var/lib/pgsql/data/postgresql.conf has not been modified and currently contains "max_connections = 100"

I would adjust your warning / critical value accordingly and also raise the max_connections to say, 200. You shouldn't see any issues with doing this.

Looking at an idle XI box, I have 78 connections running currently to SQL, so I don't think this is of any alarm.

Code: Select all

[root@localhost ~]# mysql -uroot -pnagiosxi -e "SHOW PROCESSLIST;" | wc -l
78

[TBT]

We'll make the suggested adjustments and report back if the issue remains.

Thank you,

[rkennedy]

Sounds good - I'll leave this open should any issues arise.