jangjang0401 commented on PR #1702:
URL: https://github.com/apache/cloudberry/pull/1702#issuecomment-4333079488
> > @yjhjstz @my-ship-it Regarding the vacuum_gp check failure:
> > It appears that the vacuum_gp failure is an indirect effect of this
patch. Before the fix, the standby coordinator crashed on every connection,
which caused FTS to disable synchronous replication. After the fix, however,
the standby remains healthy and synchronous replication stays enabled. As a
result, unexpected DETAIL/WARNING messages seem to appear during the fault
injection process.
> > I would appreciate it if you could confirm whether my understanding is
correct. Additionally, in this case, would it be appropriate to add SET
synchronous_commit = local before the fault injection block in vacuum_gp.sql,
or would you recommend handling this in a different way?
>
> Right direction, but I'd scope it more tightly:
>
> ```
> BEGIN;
> SET LOCAL synchronous_commit = local;
> SELECT gp_inject_fault('interconnect_stop_recv_chunk', 'interrupt', dbid)
...;
> analyze relcache_leak_in_motion;
> SELECT gp_inject_fault(... 'reset' ...);
> COMMIT;
> ```
@my-ship-it
Thank you for the suggestion! I've applied SET LOCAL synchronous_commit =
local within a BEGIN/COMMIT block as recommended.
I haven't updated vacuum_gp.out yet since I'm not entirely sure what the
exact output will look like for the commands inside the aborted transaction
state. Would it be okay to update vacuum_gp.out after checking the actual CI
output, or could you provide some guidance on the expected output?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
