Kevin Grittner <kgrittn@ymail.com> wrote:
> Andres Freund <andres@2ndquadrant.com> wrote:
>> On 2013-10-07 06:44:19 -0700, Kevin Grittner wrote:
>>
>>> Patch attached. Any objections to applying that Real Soon Now?
>>> (When, exactly is the deadline to make today's minor release
>>> cut-off?)
>>
>> Maybe it's overly careful, but I personally slightly vote for
>> applying it after the backbranch releases. The current behaviour
>> doesn't have any harsh consequences and mostly reproduceable in
>> artifical scenarios and the logic here is complex enough that we
>> might miss something.
>>
>> A day just doesn't leave much time to noticing any issues.
>
> I grant that the bug in existing production code is not likely to
> get hit very often, but it is a bug; the new isolation test shows
> the bug clearly and shows that the suggested patch fixes it.
> What tips the scales for me is that the only possible downside if
> we missed something is an occasional false positive serialization
> failure, which does not break correctness -- we try to minimize
> those for performance reasons, but the algorithm allows them and
> they currently do happen.
I am, of course, continuing to review this.
There might be a problem if someone applies this fix while any
prepared transactions are pending. Still investigating the impact
and possible fixes.
--
Kevin Grittner
EDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company
