Thread: Seeing uncommitted transactions
Is there any way to look at the database as though you were inside another sessions transaction? I've had two cases recently where this would have been somewhat useful. In one, a select into query ran for several hours and it would have been nice to see that it was running correctly. In the other an application threw a spurious BEGIN into the stream of operations, meaning that changes made weren't visible to the rest of the world until it started another transaction (which was ignored) and then committed it. Cheers, Steve
On Mon, Jun 28, 2004 at 01:33:33PM -0700, Steve Atkins wrote: > Is there any way to look at the database as though you were inside another > sessions transaction? Not currently. Maybe actually you _could_ do it with a C function, but it will require a lot of backend internal knowledge. -- Alvaro Herrera (<alvherre[a]dcc.uchile.cl>) "Y eso te lo doy firmado con mis lágrimas" (Fiebre del Loco)
Could you do this by setting debug-level higher and do log-analysis? BTW, it might be nice to give the backends a diagnostic port for "remote debugging" in similar wise as the Java Virtual Machine provides so that some degree of live-time workflow examination can happen. What do you think? Carl <|};-)> -----Original Message----- From: pgsql-general-owner@postgresql.org [mailto:pgsql-general-owner@postgresql.org] On Behalf Of Steve Atkins Sent: Monday, June 28, 2004 1:34 PM To: pgsql-general@postgresql.org Subject: [GENERAL] Seeing uncommitted transactions Is there any way to look at the database as though you were inside another sessions transaction? I've had two cases recently where this would have been somewhat useful. In one, a select into query ran for several hours and it would have been nice to see that it was running correctly. In the other an application threw a spurious BEGIN into the stream of operations, meaning that changes made weren't visible to the rest of the world until it started another transaction (which was ignored) and then committed it. Cheers, Steve ---------------------------(end of broadcast)--------------------------- TIP 7: don't forget to increase your free space map settings
"Carl E. McMillin" <carlymac@earthlink.net> writes:
> BTW, it might be nice to give the backends a diagnostic port for "remote
> debugging" in similar wise as the Java Virtual Machine provides so that some
> degree of live-time workflow examination can happen. What do you
> think?
It's called gdb ;-)
regards, tom lane