Re: Minimal logical decoding on standbys - Mailing list pgsql-hackers

On Wed, Apr 5, 2023 at 2:41 PM Drouvot, Bertrand
<bertranddrouvot.pg@gmail.com> wrote:
>
> On 4/5/23 8:59 AM, Amit Kapila wrote:
> > On Mon, Apr 3, 2023 at 12:05 PM Amit Kapila <amit.kapila16@gmail.com> wrote:
>
> > On further thinking, as such this shouldn't be a problem because all
> > the WAL records before PARAMETER_CHANGE record will have sufficient
> > information so that they can get decoded. However, with the current
> > approach, the subscriber may not even receive the valid records before
> > PARAMETER_CHANGE record. This is because startup process will
> > terminate the walsenders while invaliding the slots and after restart
> > the walsenders will exit because the corresponding slot will be an
> > invalid slot. So, it is quite possible that walsender was lagging and
> > wouldn't have sent records before the PARAMETER_CHANGE record making
> > subscriber never receive those records that it should have received.
>
> Agree that would behave that way.
>
> > I don't know whether this is what one would expect.
>
> If one change wal_level to < logical on the primary, he should at least
> know that:
>
> "
> Existing
> +     logical slots on standby also get invalidated if wal_level on primary is reduced to
> +     less than 'logical'.
> "
>
> If the doc has been read (as the quote above is coming from 0006).
>
> I think that what is missing is the "when" the slots are invalidated.
>
> Maybe we could change the doc with something among those lines instead?
>
> "
> Existing logical slots on standby also get invalidated if wal_level on primary is reduced to
> less than 'logical'. This is done as soon as the standby detects such a change in the WAL stream.
>
> It means, that for walsenders that are lagging (if any), some WAL records up to the parameter change on the
> primary won't be decoded".
>
> I don't know whether this is what one would expect but that should be less of a surprise if documented.
>
> What do you think?
>

Yeah, I think it is better to document to avoid any surprises if
nobody else sees any problem with it. BTW, another thought that
crosses my mind is that let's not invalidate the slots when the
standby startup process processes parameter_change record and rather
do it when walsender decodes the parameter_change record, if we think
that is safe. I have shared this as this crosses my mind while
thinking about this part of the patch and wanted to validate my
thoughts, we don't need to change even if the idea is valid.

minor nitpick:
+
+ /* Intentional fall through to session cancel */
+ /* FALLTHROUGH */

Do we need to repeat fall through twice in different ways?

--
With Regards,
Amit Kapila.