This patch solved a problem encountered by Evgeniy Efimkin and
Vladimir Borodin from Yandex.Mail.
and eventually deleting some of data. This testbed showed VACUUM
holding inserts for up to tenths of seconds. They claim that proposed
patch made vacuum invisible in this test.
Evgeniy, Vladimir, if I missed something or you have something to add,
please join discussion.
Yep, in our load environment the table is not so big (~ 50 GB), GIN index size is ~ 1 GB. And under our load profile we have seen 90 seconds of impossibility to do an insert into the table because of vacuuming this index. I confirm that with this patch we now don’t see any spikes of errors as it was previously.
--
May the force be with you…