I have a problem where postgresql 9.3 got stuck, and the number of postgresql processes increased from about 15 to 225 in 10 minutes.
I ran the query:
select pid, query_start, waiting, state, query from pg_stat_activity order by query_start;
But it showed mostly select statements -- all of them the same one, with a couple of joins. They are not in a waiting state but have been running for over 2 hours.
So what does this mean? Is the select query getting stuck?
Do you have a huge chunk of newly insert, not yet committed, rows? This sounds like the issue where all of the processes fight with each other over the right to check uncommitted rows in order to verify that they are actually uncommitted.
