Re: Postgres hangs during VACUUM (autocommit = false) - Mailing list pgsql-general

From Mark Coffman \(Epilogue Programmer\)
Subject Re: Postgres hangs during VACUUM (autocommit = false)
Date
Msg-id 00ee01c12a94$2978e600$e5671a18@moniker
Whole thread Raw
In response to Postgres hangs during VACUUM (autocommit = false)  ("Mark Coffman" <mark@epilogue.net>)
Responses Re: Postgres hangs during VACUUM (autocommit = false)  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-general
Thanks for the reply.  I use Perl and DBI to make my database conneciton.
In the

BEGIN {}
part of the module, I check for the persistant connection, and make one if
none is present, autocommit off
in the
END {}
block, I do a $dbh->commit();

They all still have idle in transaction
:(

----- Original Message -----
From: "Tom Lane" <tgl@sss.pgh.pa.us>
To: "Mark Coffman" <mark@epilogue.net>
Cc: <pgsql-general@postgresql.org>
Sent: Tuesday, August 21, 2001 6:28 PM
Subject: Re: [GENERAL] Postgres hangs during VACUUM (autocommit = false)


> "Mark Coffman" <mark@epilogue.net> writes:
> > is there any way for me to schedule a VACUUM
> > that won't hang?
>
> Fix your client-side code to not sit idle forever with uncommitted
> transactions.  That's a bad practice independently of whether it
> gives VACUUM problems.  The clients presumably think that whatever
> they've done so far is committed ... but guess what, it's not.
>
> > I cannot autocommit because I need transactions.
>
> Fine, but that means that you have the responsibility to commit.
> Evidently you're not doing so.
>
> FWIW, 7.2's default VACUUM will be more forgiving, but that doesn't
> make your clients' behavior a good idea.
>
> regards, tom lane
>


pgsql-general by date:

Previous
From: Mike Finn
Date:
Subject: Comparing fixed precision to floating
Next
From: Tom Lane
Date:
Subject: Re: Postgres hangs during VACUUM (autocommit = false)