Re: Rework LogicalOutputPluginWriterUpdateProgress - Mailing list pgsql-hackers

From Amit Kapila
Subject Re: Rework LogicalOutputPluginWriterUpdateProgress
Date
Msg-id CAA4eK1J4ENF6yicvgHVNbEwxeGpeRYGu6EmS8mgpjSfB4wFAkA@mail.gmail.com
Whole thread Raw
In response to RE: Rework LogicalOutputPluginWriterUpdateProgress  ("wangw.fnst@fujitsu.com" <wangw.fnst@fujitsu.com>)
Responses RE: Rework LogicalOutputPluginWriterUpdateProgress
List pgsql-hackers
On Wed, Mar 8, 2023 at 8:24 AM wangw.fnst@fujitsu.com
<wangw.fnst@fujitsu.com> wrote:
>
> Attach the new patch.
>

I think this combines multiple improvements in one patch. We can
consider all of them together or maybe it would be better to split
some of those. Do we think it makes sense to split some of the
improvements? I could think of below:

1. Remove SyncRepRequested() check from WalSndUpdateProgress().
2. Add check of wal_sender_timeout > 0 in WalSndUpdateProgress() and
any other similar place.
3. Change the name of ProcessPendingWrites() to WalSndSendPending().
4. Change WalSndUpdateProgress() to WalSndUpdateProgressAndKeepalive().
5. The remaining patch.

Now, for (1), we can consider backpatching but I am not sure if it is
worth it because in the worst case, we will miss sending a keepalive.
For (4), it is not clear to me that we have a complete agreement on
the new name. Andres, do you have an opinion on the new name used in
the patch?

If we agree that we don't need to backpatch for (1) and the new name
for (4) is reasonable then we can commit 1-4 as one patch and then
look at the remaining patch.

Thoughts?

--
With Regards,
Amit Kapila.



pgsql-hackers by date:

Previous
From: Masahiko Sawada
Date:
Subject: Re: Should vacuum process config file reload more often
Next
From: Michael Paquier
Date:
Subject: Re: Combine pg_walinspect till_end_of_wal functions with others