On 25 March 2014 19:59, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Robert Haas <robertmhaas@gmail.com> writes:
>> This commit message doesn't give any indication of what the
>> consequences of this oversight are. I think that's generally a good
>> thing to include.
>
> Yes, please. We (usually Bruce or I) make release notes based on the
> commit logs. Please be sure your commit message includes enough info
> to write a useful release note about the change.
I don't think there are any consequences. All it seems to do is slow
down recovery by an XLogFlush(), which admitedly isn't much, so I'm
not asking to revert either.
The WAL record concerned is written immediately before the shutdown
checkpoint at the end of recovery, which then definitely flushes WAL.
That is very definitely there before 9.3.
If there isn't a shutdown checkpoint then we simply begin recovery at
the last restartpoint, and then rewrite the parameter change, so even
in 9.3 I don't see a problem.
--
Simon Riggs http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services