On 11 June 2015 at 17:34, Robert DiFalco <robert.difalco@gmail.com> wrote:
> I want to make sure I understand the repercussions of this before making it
> a global setting.
>
> As far as I can tell this will put data/referential integrity at risk. It
> only means that there is a period of time (maybe 600 msecs) between when a
> commit occurs and when that data is safe in the case of a server crash.
There should be no risk to referential integrity. All it means is
that the changes won't definitely be in the WAL on disk at the time of
a command reporting that it has successfully completed. If a crash
were to occur, any changes that would depend on such a command
wouldn't be committed either, so the database should remain in a
consistent state.
So this only risks loss of changes over a short period of time, not
risk of corruption or loss of integrity.
Regards
Thom