Tom Lane wrote:
Karl Denninger <karl@denninger.net> writes:
But... .shouldn't autovacuum prevent this? Is there some way to look in
a log somewhere and see if and when the autovacuum is being run - and on
what?
There's no log messages (at the default log verbosity anyway). But you
could look into the pg_stat views for the last vacuum time for each table.
regards, tom lane
It looks like it IS being run.
Now the problem is, what's going on? Maybe running out of fsm_map entries? Hmnmmmmm..... vacuum full fixes it, which I assume reclaims those, yes? I wonder if increasing that quite radically (I have lots of server RAM and SYSV SHM has been cranked wide) would take care of it....
-- Karl