On Mon, 29 Sep 2008 19:06:46 -0400 (EDT)
Bruce Momjian <bruce@momjian.us> wrote:
> I agree with Tom; either we make the pg_start_backup() checkpoint
> immediate or leave the behavior unchanged.
>
> Personally I think immediate makes more sense because issuing
> pg_start_backup() seems like it should behave like a manual CHECKPOINT
> command.
>
I am not actually that concerned that it takes a long time now that I
know why. I am more concerned that it isn't obvious why it takes so
long. Something like this would be more than sufficient:
elog(NOTICE, "pg_start_backup called, issuing CHECKPOINT");
RequestCheckpoint(CHECKPOINT_FORCE | CHECKPOINT_WAIT);
Sincerely,
Joshua D. Drake
--
The PostgreSQL Company since 1997: http://www.commandprompt.com/
PostgreSQL Community Conference: http://www.postgresqlconference.org/
United States PostgreSQL Association: http://www.postgresql.us/