pg_stats_subscription_stats order of the '*_count' columns - Mailing list pgsql-hackers

From Peter Smith
Subject pg_stats_subscription_stats order of the '*_count' columns
Date
Msg-id CAHut+PvbOw90wgGF4aV1HyYtX=6pjWc+pn8_fep7L=aLXwXkqg@mail.gmail.com
Whole thread Raw
List pgsql-hackers
Hi,

While reviewing another thread I was looking at the view
'pg_stats_subscription_stats' view. In particular, I was looking at
the order of the "*_count" columns of that view.

IMO there is an intuitive/natural ordering for the logical replication
operations (LR) being counted. For example, LR "initial tablesync"
always comes before LR "apply".

I propose that the columns of the view should also be in this same
intuitive order: Specifically, "sync_error_count" should come before
"apply_error_count" (left-to-right in view, top-to-bottom in docs).

Currently, they are not arranged that way.

The view today has only 2 count columns in HEAD, so this proposal
seems trivial, but there is another patch [2] soon to be pushed, which
will add more conflict count columns. As the number of columns
increases IMO it becomes more important that each column is where you
would intuitively expect to find it.

~

Changes would be needed in several places:
- docs (doc/src/sgml/monitoring.sgml)
- function pg_stat_get_subscription_stats (pg_proc.dat)
- view pg_stat_subscription_stats (src/backend/catalog/system_views.sql)
- TAP test SELECTs (test/subscription/t/026_stats.pl)

~

Thoughts?

======
[1] docs - https://www.postgresql.org/docs/devel/monitoring-stats.html#MONITORING-PG-STAT-SUBSCRIPTION-STATS
[2] stats for conflicts -

https://www.postgresql.org/message-id/flat/OS0PR01MB57160A07BD575773045FC214948F2%40OS0PR01MB5716.jpnprd01.prod.outlook.com

Kind Regards,
Peter Smith.
Fujitsu Australia



pgsql-hackers by date:

Previous
From: Thomas Munro
Date:
Subject: Windows socket problems, interesting connection to AIO
Next
From: Peter Smith
Date:
Subject: Re: Collect statistics about conflicts in logical replication