Thread: [PERFORM] Postgres Dump - Creating index never stops
Hello Experts,
we have created a postgres dump using this command: pg_dump pixabay | gzip > pixabay.gz
After restarting postgres (even with a new cluster) and creating a new database, postgres is hanging while extracting the dump: gunzip -c pixabay.gz | psql pixabay
The log file shows that the autovacuum task is running (almost) endless...
2017-07-12 18:05:50.822 CEST [19586] hans@pixabay LOG: duration: 1594.319 ms statement: CREATE INDEX photos_download_photo_id ON photos_download USING btree (photo_id);
2017-07-12 18:05:52.340 CEST [19586] hans@pixabay LOG: duration: 1517.955 ms statement: CREATE INDEX photos_download_user_id ON photos_download USING btree (user_id);
2017-07-12 18:05:52.355 CEST [19586] hans@pixabay LOG: duration: 14.954 ms statement: CREATE INDEX photos_editorschoice_created ON photos_editorschoice USING btree (created);
2017-07-12 18:05:52.367 CEST [19586] hans@pixabay LOG: duration: 11.609 ms statement: CREATE INDEX photos_indexphoto_created ON photos_indexphoto USING btree (created);
2017-07-12 20:34:58.943 CEST [19626] ERROR: canceling autovacuum task
2017-07-12 20:34:58.943 CEST [19626] CONTEXT: automatic analyze of table "pixabay.public.photos_photo"
2017-07-12 20:34:59.942 CEST [19586] hans@pixabay LOG: duration: 8947575.013 ms statement: CREATE INDEX photos_photo_approved_by_id ON photos_photo USING btree (approved_by_id);
2017-07-12 20:35:00.471 CEST [19586] hans@pixabay LOG: duration: 528.777 ms statement: CREATE INDEX photos_photo_approved_date ON photos_photo USING btree (approved_date);
What could cause this problem or how can we debug it?
We are running Postgres 9.4 / Ubuntu 16.04
Thanks, Hans
Hans Braxmeier <hans.braxmeier@outlook.com> writes: > After restarting postgres (even with a new cluster) and creating a new database, postgres is hanging while extracting thedump: gunzip -c pixabay.gz | psql pixabay > The log file shows that the autovacuum task is running (almost) endless... > 2017-07-12 18:05:52.367 CEST [19586] hans@pixabay LOG: duration: 11.609 ms statement: CREATE INDEX photos_indexphoto_createdON photos_indexphoto USING btree (created); > 2017-07-12 20:34:58.943 CEST [19626] ERROR: canceling autovacuum task > 2017-07-12 20:34:58.943 CEST [19626] CONTEXT: automatic analyze of table "pixabay.public.photos_photo" > 2017-07-12 20:34:59.942 CEST [19586] hans@pixabay LOG: duration: 8947575.013 ms statement: CREATE INDEX photos_photo_approved_by_idON photos_photo USING btree (approved_by_id); What that looks like is it took the system an unusually long time to notice that it needed to cancel the autovacuum to avoid a deadlock with the CREATE INDEX. Was either process consuming a noticeable amount of CPU during that interval? Do you have deadlock_timeout set higher than the default 1s? regards, tom lane