[pgpool-general: 6987] Best practices for monitoring pgpool?

[Martin Goodson]

Hello.

We have a pgpool configuration consisting of three pgpool servers 
running watchdog with a delegate IP to allow the application servers to 
access the primary pgpool. The backend consists of two PostgreSQL 10.5 
servers running native physical replication.

The monitoring for this is rather rudimentary - we're regularly running 
pcp_watchdog_info to monitor the state of the cluster on each server, 
and combining show pool_nodes and pcp_node_info to get the state of the 
backends. In the event of a chance of state (e.g. a server changes from 
'MASTER' to 'STANDBY', a server changes state to LOST, we lose quorum, 
etc) our monitoring team receive an e-mail. Things like running out of 
disk, CPU hitting 100%, running out of memory, server crashing etc is 
handled by a separate set of server monitoring. That's already covered, 
at this point it's pgpool we're interested in.

We've now been asked to provide a feed into kibana consisting of the 
info from above commands plus 'output from any other command that might 
prove useful', plus any 'useful messages' that might arise in the pgpool 
log. In short, our kibana team wants a log of 'everything vital and 
nothing else' ...

So I was just wondering what people are looking for in the pgpool logs, 
commands that they use to monitor pgpool, etc. Is there any 
recommendation for what commands are most useful for monitoring, what 
messages need to be addressed and which can be safely ignored, etc?

Bit of a long shot, I know, since there are so many variables in this 
sort of thing, but I was asked to see if there was 'any consensus in the 
pgpool community' so here I am :)

Many thanks!

M.

-- 
Martin Goodson

In bed above we're deep asleep,
While greater love lies further deep.
This dream must end, the world must know,
We all depend on the beast below.