From: | "Boris S(dot)" <boris(at)splet(dot)net> |
---|---|
To: | pgsql-admin(at)postgresql(dot)org |
Subject: | High CPU usage from a vacuum process but does nothing |
Date: | 2017-10-04 17:28:35 |
Message-ID: | ca07e0e6-d51b-720b-337f-e5f5e053e533@infosplet.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-admin |
After migrating the server to a bigger machine, I'm noticing an odd
behaviour. If I begin vacuuming manually, it starts OK:
<snip>
0 pages are entirely empty.
CPU 0.00s/0.00u sec elapsed 0.00 sec.
INFO: analyzing "public.table024484"
INFO: "table024484": scanned 111123 of 111123 pages, containing 4909226
live rows and 0 dead rows; 150000 rows in sample, 4909226 estimated
total rows
But then it gets stuck indefinitely with high CPU usage and command
never exits. kill -TERM isn't helping, process stays on with ~99% CPU
usage. Stracing this process gets me just:
strace: Process 102011 attached
strace: [ Process PID=102011 runs in x32 mode. ]
Which is an odd message, all other processes give normal strace output.
latest pgsql 9.5 (same on old and new machine)
Ubuntu 16.04
Values in "cat /proc/102011/io" stay the same.
Autovacuum and other processes, queries aren't affected and can be
straced normally. Disk IO is low (checked with iostat)
Any ideas on what else to try?
From | Date | Subject | |
---|---|---|---|
Next Message | ghiureai | 2017-10-04 19:01:32 | streaming replication corruption Q's |
Previous Message | Alvaro Aguayo Garcia-Rada | 2017-10-04 16:00:26 | Re: replication issue |