Hi,
On 2020-11-19 09:45:50 -0700, laurent.mignon@acsone.eu wrote:
> We also encounter this issue on our postgresql12 database. Disabling jit
> inlining seems to reduce drastically the memory leak. But after 4 days
> without restart (compared to the 3 restart / day before the change) it
> seems that we still have a leak. The RAM is full and the swap almost
> full..
In that case it is a separate issue from Kurt's. Could you follow the
instructions from
https://www.postgresql.org/message-id/1999972.1604957673%40sss.pgh.pa.us
and
https://www.postgresql.org/message-id/20201110043127.sfkyvvjqy7x3er5k%40alap3.anarazel.de
providing the output here?
> I hope that a fix will land quickly since this issue has a huge impact on
> our business.
It's unfortunately a nontrivial issue - I have a prototype for a fix,
but it's complicated and not yet complete. And unfortunately I've since
injured my hand and am now typing very slowly...
Btw, I'll be more likely to see your reply promptly if you CC me in your
reply.
Greetings,
Andres Freund