Re: race condition in sync rep

Simon Riggs <simon(at)2ndQuadrant(dot)com> writes:
> On Sat, Mar 26, 2011 at 4:24 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> Well, in that case, it should be on the open-items list. If the system
>> is still behaving that way, it's a bug.

> Is it? Sync rep requires fsync on the standby. If you then explicitly
> turn off fsync on the standby then it has a performance impact, as
> documented.

No, that's utter nonsense. If the system is behaving that way, then
it's a bug. If you don't think it's a bug, then you misunderstand what
the fsync GUC is supposed to do. What fsync=off is supposed to do is
cause every attempted fsync to succeed instantly. It is *not possible*
for that to result in a performance slowdown compared to fsyncs that
take nonzero time. Unless someone's broken it. The documented behavior
is broken. Period.

regards, tom lane