On 02/25/2014 08:06 AM, Tom Lane wrote:
> Well, it's three months later, and none of the people who were complaining
> so vociferously in this thread seem to have bothered to test the proposed
> solution.
Sorry about that. The client lost interest once they had a workaround
(fixing the hanging transaction), and I don't have direct access to
their test workload -- nor was I able to reproduce the issue on a purely
synthetic workload.
> However, over at
> http://www.postgresql.org/message-id/CAFj8pRDHyAK_2JHSVKZ5YQNGQmFGVcJKcpBXhFaS=vSSCH-vNw@mail.gmail.com
> Pavel did test it and reported that it successfully alleviates his
> real-world problem. So I'm now inclined to commit this. Objections?
None from me.
--
Josh Berkus
PostgreSQL Experts Inc.
http://pgexperts.com