On Tue, Oct 23, 2018 at 20:23:14 -0500,
Bruno Wolff III <br...@wolff.to> wrote:
While reloading a database cluster to move from 10.5 to 11, I'm
getting out of memory crashes that I did see when doing reloads on pg
10.
The statement flagged in the log is this:
2018-10-23 16:44:34.815 CDT [126839] STATEMENT: ALTER TABLE ONLY
public.iplocation
ADD CONSTRAINT overlap EXCLUDE USING spgist (network WITH &&);
iplocation has 4398722 rows.
I'm trying to reproduce this on my desktop but so far it isn't producing
the same issue. The database is still loading after 9 hours, but it looks
like it got past the point where the problem index was created. (I'm not
sure how to check if the index has really finished being created, but \d
shows it as existing.)
I'll know better tomorrow.
My workstation is using the Fedora version of postgresql 11 which might
have some relevant difference from the pgdg version for rhel7. I still
have a number of things I can try, but they might take significant time
to run and I might not get a reasonable test case for a while.
