Re: Autovacuum Daemon Disrupting dropdb?

"Thomas F. O'Connell" <tfo(at)sitening(dot)com> writes:
> On Mar 11, 2006, at 4:13 PM, Tom Lane wrote:
>> For a "real" solution, perhaps DROP DATABASE could somehow look to
>> determine if there's an autovac daemon active in the target database,
>> and if so send it a SIGINT and wait for it to go away.

> In general, it also seems like a --force option or something similar
> would be reasonable for dropdb because the state of the database in
> terms of user activity wouldn't seem to matter a whole lot if the
> intent is to drop it.

... except to the processes connected to it.

If we trusted selective SIGTERM we could imagine sending that to
non-autovac processes connected to the target database, but we don't
really. In any case, killing a database that has active users seems
like a pretty large-caliber foot-gun to me; that condition suggests
*very* strongly that the database is not so idle as all that.

regards, tom lane