Re: WalSndWakeup() and synchronous_commit=off

On Friday, May 11, 2012 08:36:24 PM Tom Lane wrote:
> Robert Haas <robertmhaas(at)gmail(dot)com> writes:
> > That definitely doesn't seem ideal - a lot of things can pile up
> > behind WALWriteLock. I'm not sure how big a problem it would be in
> > practice, but we generally make a practice of avoiding sending signals
> > while holding LWLocks whenever possible...
>
> There's a good reason for that, which is that the scheduler might well
> decide to go run the wakened process instead of you. Admittedly this
> tends to not be a problem on machines with $bignum CPUs, but on
> single-CPU machines I've seen it happen a lot.
>
> Refactoring so that the signal is sent only after lock release seems
> like a good idea to me.
Will send a patch lateron, duplication seems to be manageable.

Andres