We have a 8 core ESXi. We have a few virtual servers on it, one of which is our database (virtual) server. We have 2 cores provisioned for that db server.
I have set:
set max_parallel_workers_per_gather to 2
And max_worker_processes
was already at 8.
I followed this link to setup a loadavg
foreign data table
:
https://aaronparecki.com/2015/02/19/8/monitoring-cpu-memory-usage-from-postgres
So I can do this:
SELECT * FROM loadavg;
I read these links, and that's about what I know on this:
http://blog.scoutapp.com/articles/2009/07/31/understanding-load-averages
Understanding load average vs. cpu usage
How do I confirm that my queries are benefiting from the max_parallel_workers_per_gather
setting of 2 (instead of 0)?
Will the max_parallel_workers_per_gather
setting affect what I see for my load averages (in the loadavg
table I setup)?
I'm seeing averages near 4 and scheduled is sometimes 4/190. Seems alarmingly high given the links I have read above.
Bonus if you provide background that goes beyond my question. Thanks!
It is thin provisioned if that matters.