Re: Separate GUC for replication origins - Mailing list pgsql-hackers

From Peter Eisentraut
Subject Re: Separate GUC for replication origins
Date
Msg-id f8ca5699-8ea8-4c75-946a-30154301aa73@eisentraut.org
Whole thread Raw
In response to Re: Separate GUC for replication origins  (Peter Eisentraut <peter@eisentraut.org>)
List pgsql-hackers
On 11.02.25 21:25, Euler Taveira wrote:
> Here is another patch that only changes the GUC name to
> max_replication_origin_sessions.

I think the naming and description of this is still confusing.

What does this name mean?  There is (I think) no such thing as a 
"replication origin session".  So why would there be a maximum of those?

The description in the documentation, after the patch, says "Specifies 
how many replication origins (see Chapter 48) can be tracked 
simultaneously".  But Chapter 48 does not say anything about what it 
means for a replication slot to be tracked.  The only use of the word 
tracked in that chapter is to say that replication slots do the tracking 
of replication progress.

Both of these are confusing independently, but moreover we are now using 
two different words ("sessions" and "tracked") for apparently the same 
thing, but neither of which is adequately documented in those terms 
anywhere else.

I agree that the originally proposed name max_replication_origins is not 
good, because you can "create" (using pg_replication_origin_create()) 
more than the configured maximum.  What is the term for what the setting 
actually controls?  How many are "active"?  "In use"?  Per session?  etc.




pgsql-hackers by date:

Previous
From: Dean Rasheed
Date:
Subject: Re: Wrong results with subquery pullup and grouping sets
Next
From: Matthias van de Meent
Date:
Subject: Re: Incorrect result of bitmap heap scan.