Forum OpenACS Q&A: Postmaster is eating my CPU

Collapse
1: Postmaster is eating my CPU
Posted by James Thornton on 09/03/01 08:52 PM
Postmaster is eating my CPU -- see ps and top output at http://jamesthornton.com/misc/postgres.txt (it wrapped too much when I tried to post it to the bboard).

As you can see from the ps output, there are several INSERT statements -- these return after restarting Postgres and even rebooting the system. I checked the system log for that server, and there are only ~30 INSERTS over the last ~12 hours (all INSERT into referer_log). Futhermore, I haven't been running any INSERT statements from psql, and no one else has access to this system.

Yesterday, I ran "vacuum analyze" for the first time in a long time -- could that have caused this situation?

Collapse
2: Response to Postmaster is eating my CPU (response to 1)
Posted by Don Baccus on 09/03/01 09:31 PM
This is very strange ... my guess is that for some reason a lock is being held persistently and your processes are spinning on it.  This should never (cough) happen.

This is one for the PG hackers group, I think - Tom Lane's more likely
to be able to give you help here than any of us.

Collapse
3: Response to Postmaster is eating my CPU (response to 1)
Posted by good bye on 10/12/01 08:51 AM
did you ever get any resolution on this? the same thing just
happened to me today. however, it looks like a SELECT is
causing postgres to spin. the only thing i really did to the
database in the past 24 hours is rebuild some views. and, the
site goes live tomorrow. perfect timing! 😟
Collapse
4: Response to Postmaster is eating my CPU (response to 1)
Posted by good bye on 10/12/01 01:57 PM
I figured this out in my case. I had altered a table by renaming a
column. The view that my page was using used that particular
column to determine what rows were a part of the view... Which
ended up causing some serious weirdness. so, just a word of
warning: dont do what i did.