> > This is also interesting. I just allowed connections to template0
> > for the express purpose of vacuuming it, did a full vac on
> > template0, and that did NOT clear up the hanging clogs.
>
> Was it a VACUUM FREEZE followed by a CHECKPOINT? Did you also cover
> that new database?
Yes, yes and yes, actually. Unless I did it wrong, which is always a strong possibility. Especially early in the
morning...
pg statement: update pg_database set datallowconn=true where datname='template0';
shell: vacuumdb -f -v -z -F template0
pg statement: checkpoint;
pg statement: update pg_database set datallowconn=true where datname='template0';
Like I said, I'm beginning to believe that something is rotten wrt my Saturday vacs...