[Spacewalk-list] spacewalk eats up all connections to postgres

[Jonathan Scott]

On Wed, Dec 12, 2012 at 12:32 PM, Paul Robert Marino <prmarino1 at gmail.com>wrote:

> I was the first one to point out the jdbc driver
> it does help significantly but i never could find any thing in the
> change log to explain why it fixes the issue
> its not a 100% fix but you will notice a 95% reduction in idle
> connections and it will be relatively stable give or take 5 idle
> connections.
>
>
I remember you doing so, and I was one of the ones who started testing as
soon as you posted. Unfortunately, it didn't do anything to help in my
situation as far as I could tell (perhaps slowing down but never reducing).
I still had stacking idle transactions necessitating nightly restarts. My
idle transaction count would quadruple every few hours until postgresql
stopped accepting connections. Adjusting the connection count only delayed
the inevitable.

Applying the patch to python-psycopg2 is the only thing that has thus far
proven to provide an actual fix to my issues. I have been running solid for
days, never breaking more than 10 idle transactions (consistently running
around 3-4), and not needing a full application recycle (and ipcs flush of
apache processes) yet.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/spacewalk-list/attachments/20121212/83d06b2b/attachment.htm>