Re: pendingOps table is not cleared with fsync=off

On Mon, Aug 10, 2020 at 02:50:26PM -0400, Tom Lane wrote:
>
> Shawn Debnath <sdn(at)amazon(dot)com> writes:
> > Good catch. Question is, are the users aware of the requirement to do a
> > manual fsync if they flip the fsync GUC off and then on? Should we do
> > this on their behalf to make a good faith attempt to ensure things are
> > flushed properly via an assign hook?
>
> No. Or at least, expecting that you can do that from an assign hook
> is impossibly wrong-headed. GUC assign hooks can't have failure modes.

Okay agree, will remind myself to drink more coffee next time.

If we think a fsync should be performed in this case, assign hook
could set a value to indicate parameter was reset via SIGHUP. Next call
to ProcessSyncRequests() could check for this, do a fsync prior to
absorbing the newly submitted sync requests, and reset the flag.
fsync_pgdata() comes to mind to be inclusive.

If folks are not inclined to do the fsync, the change is good as is.

--
Shawn Debnath
Amazon Web Services (AWS)