Re: Unhappy about API changes in the no-fsm-for-small-rels patch

On Tue, Apr 30, 2019 at 6:06 PM Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> wrote:
>
> On Tue, Apr 30, 2019 at 2:24 PM Dilip Kumar <dilipbalaut(at)gmail(dot)com> wrote:
> >
> > insert into atacc1 values (21, 22, 23);
> > +ERROR: could not read block 0 in file "base/16384/31379": read only
> > 0 of 8192 bytes
> >
> > I have analysed this failure. Seems that we have not reset the
> > rel->fsm_local_map while truncating the relation pages by vacuum
> > (lazy_truncate_heap). So when next time while accessing it we are
> > getting the error. I think we need a mechanism to invalidate this
> > when we truncate the relation pages. I am not sure whether we should
> > invalidate the relcache entry here or just reset the
> > rel->fsm_local_map?
> >
>
> Thanks, this appears to be the missing case where we need to
> invalidate the cache. So, as discussed above if we issue invalidation
> call (in RecordPageWithFreeSpace) when the page becomes empty, then we
> shouldn't encounter this. John, can we try this out and see if the
> failure goes away?

I added a clear/inval call in RecordPageWithFreeSpace and the failure
goes away. Thanks for the analysis, Dilip!

--
John Naylor https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services