Horton

Re: dropdb --force

Filip Rembiałkowski <filip.rembialkowski@gmail.com>

View thread View on postgr.es
From: Filip Rembiałkowski <filip.rembialkowski@gmail.com>
To: Andres Freund <andres@anarazel.de>
Cc: Thomas Munro <thomas.munro@gmail.com>, Marti Raudsepp <marti@juffo.org>, Pavel Stehule <pavel.stehule@gmail.com>, Pgsql Hackers <pgsql-hackers@postgresql.org>
Date: 2019-04-01T11:26:06Z
Lists: pgsql-hackers 
On 31.03.2019, 04:35 Andres Freund <andres@anarazel.de> wrote:

>
> >  bool
> > -CountOtherDBBackends(Oid databaseId, int *nbackends, int *nprepared)
> > +CountOtherDBBackends(Oid databaseId, int *nbackends, int *nprepared, bool force_terminate)
> >  {
>
> That doesn't seem like a decent API to me.

Only excuse is that naming was already a bit off, as the function
includes killing autovacuum workers.

Please advise what would be a good approach to improve it. I would
propose something like:


bool CountOtherDBBackends(Oid databaseId, int *nbackends, int *nprepared);
// make it actually do what the name announces - only the count, no
side effects.

bool KillDBBackends(Oid databaseId, bool killAutovacuum, bool killBackends);
// try to kill off all the backends, return false when there are still any.



Also, there is a question - should the FORCE option rollback prepared
transactions?

Thanks!

Commits

  1. Add tests for '-f' option in dropdb utility.

  2. Move pump_until to TestLib.pm.

  3. Add the support for '-f' option in dropdb utility.

  4. Introduce the 'force' option for the Drop Database command.

  5. Improve CREATE/DROP/RENAME DATABASE so that when failing because the source