> > It removes the need to disable fsync to get best performance!
>
> -F performance is still better, only the difference is not so big as before.
Well, when "checkpoint seek in logs" will be implemented difference
will be the same - lost consistency.
> > Since there is a fundamental recovery problem if the WAL file
> > disappears, then perhaps we should have a workaround which can ignore
> > the requirement for that file on startup? Or maybe we do already?
> > Vadim??
>
> This was discussed, but iirc not yet implemented.
Yes & yes.
> > Also, could the "-F" option be disabled now that WAL is enabled? Or is
> > there still some reason to encourage/allow folks to use it?
I've used it when testing btree runtime recovery to increase concurrence.
> I use it, since I restore after a system crash (which never happens).
> I think all that is probably missing in -F mode is probably 2-3 fsyncs
> during checkpoint. One for the xlog, and one for pg_control (maybe also pg_log).
> All other fsyncs are only to not buffer transactions.
Probably we could just force fsync during checkpoint, for the moment.
Thanks to all for help!
Vadim