Hi,
On 1/17/23 12:23 PM, Alvaro Herrera wrote:
On 2023-Jan-17, Drouvot, Bertrand wrote:
The idea has been raised in [1], where we are adding more calls to
wait_for_catchup() in 'replay' mode.
This seems mostly useless as presented. Maybe if you're able to reduce
the noise on the second argument it would be worth something -- namely,
if the wrapper function receives a node instead of an LSN: perhaps
wait_for_replay_catchup() would use the flush LSN from the given node,
wait_for_write_catchup() would use the write LSN, and
wait_for_sent_catchup() would use the insert LSN. (I didn't check in
your patch if there are callsites that do something else). This would
in several cases let you also remove the line with the assignment of
appropriate LSN to a separate variable. If you did it that way, maybe
the code would become a tiny bit smaller overall.
Thanks for looking at it!
The current calls are done that way:
wait_for_replay_catchup called:
- 8 times with write LSN as an argument
- 1 time with insert LSN as an argument
- 16 times with flush LSN as an argument
wait_for_write_catchup called:
- 5 times with write LSN as an argument
So it looks like that providing a node as a second argument
would not help for the wait_for_replay_catchup() case.
Worth to use the node as an argument for wait_for_write_catchup()? (though it
would be
weird to have different types of arguments between wait_for_replay_catchup()
and wait_for_write_catchup()).
Regards,
--
Bertrand Drouvot
PostgreSQL Contributors Team
RDS Open Source Databases
Amazon Web Services: https://aws.amazon.com
