The server has been experiencing high loads periodically since a rebuild hang. AstonJ thinks they’ve found the trigger: running complex queries, specifically in ACP > Reports > Top Traffic Sources with a large date range. pfaffman suggests increasing RAM for Postgres, which AstonJ tries by upping db_shared_buffers to 10GB. However, this doesn’t resolve the issue. AstonJ suspects a version change around the time of the rebuild hang is the cause. pfaffman recommends further increasing db_work_mem to at least 100MB.
Ever since a rebuild hanged we’ve been experiencing high loads periodically with no obvious cause (other than it started after that upgrade/hang).
Well tonight I think I was able to manually trigger it - by going to the ACP > Reports > Top Traffic Sources then setting the dates from the beginning of the year to today. So it appears big/complex queries may be triggering it - anyone else experiencing the same or also able to trigger it?
You might need more ram or to give more ram to postgres.
Do you mean in the container under db_shared_buffers: "4096MB"? It says 25% of total memory (server has 64GB) so Ill try upping this to 10GB.
That’s the only performance related setting I can recall… are there any others we should be aware of or routinely checking/tweaking?
Edit: just looked and there’s also:
## can improve sorting performance, but adds memory usage per-connection
#db_work_mem: "40MB"
Any threads that discuss this anywhere? (Or do you or anyone else have any thoughts?)
Increased to 10GB and tried again - same issue tho.
These problems started after that rebuild hang so I suspect it is something to do with some kind of change from a version around that time 
I’d bump that to at least 100MB