On 2/15/21 1:50 PM, Joe Abbate wrote:
On 15/2/21 16:29, Adrian Klaver wrote:
On 2/15/21 1:15 PM, Joe Abbate wrote:
We've been experiencing PG server process crashes about every other
week on a mostly read only website (except for a single insert/update
on page access). Typical log entries look like
LOG: checkpointer process (PID 11200) was terminated by signal 9:
Killed
LOG: terminating any other active server processes
Have you looked at the system logs to see if the OOM killer is involved?
No, I hadn't, but now I see that of the past three crashes, four
postgres processes were OOM victims. I guess I have to read the OOM
articles I've been skipping (read one a long time ago). If you have any
specific suggestions, let me know.
There are suggestions here:
https://www.postgresql.org/docs/12/kernel-resources.html#LINUX-MEMORY-OVERCOMMIT
It probably would not hurt to figure why this seemed to happen with the
Python -> Go switch. Even if you can get the OOM not to kick in, the
fact that it was kicking in would indicate you now have memory hungry
processes that did not exist before. The questions being is this
strictly due to the language/framework change, natural growth of data
set being worked with, or something else?
Thanks Adrian,
Joe
--
Adrian Klaver
adrian.klaver@xxxxxxxxxxx
