Re: Improving connection scalability: GetSnapshotData()

On 4/8/20 8:59 AM, Alexander Korotkov wrote:
> On Wed, Apr 8, 2020 at 3:43 PM Andres Freund <andres@anarazel.de> wrote:
>> Realistically it still 2-3 hours of proof-reading.
>>
>> This makes me sad :(
>
> Can we ask RMT to extend feature freeze for this particular patchset?
> I think it's reasonable assuming extreme importance of this patchset.

One of the features of RMT responsibilities[1] is to be "hands off" as
much as possible, so perhaps a reverse ask: how would people feel about
this patch going into PG13, knowing that the commit would come after the
feature freeze date?

My 2¢, with RMT hat off:

As mentioned earlier[2], we know that connection scalability is a major
pain point with PostgreSQL and any effort that can help alleviate that
is a huge win, even in incremental gains. Andres et al experimentation
show that this is more than incremental gains, and will certainly make a
huge difference in people's PostgreSQL experience. It is one of those
features where you can "plug in and win" -- you get a performance
benefit just by upgrading. That is not insignificant.

However, I also want to ensure that we are fair: in the past there have
also been other patches that have been "oh-so-close" to commit before
feature freeze but have not made it in (an example escapes me at the
moment). Therefore, we really need to have consensus among ourselves
that the right decision is to allow this to go in after feature freeze.

Did this come in (very) late into the development cycle? Yes, and I
think normally that's enough to give cause for pause. But I could also
argue that Andres is fixing a "bug" with PostgreSQL (probably several
bugs ;) with PostgreSQL -- and perhaps the fixes can't be backpatched
per se, but they do improve the overall stability and usability of
PostgreSQL and it'd be a shame if we have to wait on them.

Lastly, with the ongoing world events, perhaps time that could have been
dedicated to this and other patches likely affected their completion. I
know most things in my life take way longer than they used to (e.g.
taking out the trash/recycles has gone from a 15s to 240s routine). The
same could be said about other patches as well, but this one has a far
greater impact (a double-edged sword, of course) given it's a feature
that everyone uses in PostgreSQL ;)

So with my RMT hat off, I say +1 to allowing this post feature freeze,
though within a reasonable window.

My 2¢, with RMT hat on:

I believe in[2] I outlined a way a path for including the patch even at
this stage in the game. If it is indeed committed, I think we
immediately put it on the "Recheck a mid-Beta" list. I know it's not as
trivial to change as something like "Determine if jit="on" by default"
(not picking on Andres, I just remember that example from RMT 11), but
it at least provides a notable reminder that we need to ensure we test
this thoroughly, and point people to really hammer it during beta.

So with my RMT hat on, I say +0 but with a ;)

Thanks,

Jonathan

[1] https://wiki.postgresql.org/wiki/Release_Management_Team#History
[2]
https://www.postgresql.org/message-id/6be8c321-68ea-a865-d8d0-50a3af616463%40postgresql.org