RE: Conflict detection for update_deleted in logical replication

On Wed, Jun 11, 2025 at 5:34 PM shveta malik wrote:
>
> On Tue, Jun 10, 2025 at 11:55 AM Zhijie Hou (Fujitsu)
> <houzj(dot)fnst(at)fujitsu(dot)com> wrote:
> >
> > Here is the V35 patch set which includes the following changes:
> >
>
> Thank You for the patches, few comments:

Thanks for the comments!

>
> 1)
> compute_min_nonremovable_xid:
>
> + /*
> + * Stop advancing xmin if an invalid non-removable transaction ID is
> + * found, otherwise update xmin.
> + */
> + if (!TransactionIdIsValid(nonremovable_xid))
> + *can_advance_xmin = false;
>
> IMO, we can not have invalid nonremovable_xid here. It may be a possibility in
> a later patch where we do stop-conflict-detection for a worker and resets
> nonremovable_xid to Invalid. But in patch003, should this rather be an Assert
> as we want to ensure that worker's oldest is set to slot's-xmin by this time.

Agreed. I have removed the check and added an Assert.

>
> 2)
> postgres=# create subscription sub1 connection 'dbname=postgres
> host=localhost user=shveta port=5433' publication pub1 WITH
> (retain_conflict_info = true);
> NOTICE: deleted rows will continue to accumulate for detecting conflicts
> until the subscription is enabled
> NOTICE: created replication slot "sub1" on publisher CREATE
> SUBSCRIPTION
>
> postgres=# alter subscription sub2 enable;
> NOTICE: deleted rows will continue to accumulate for detecting conflicts
> until the subscription is enabled
>
> We should not have this 'NOTICE: deleted rows' in above 2 commands.

Agreed, It was a miss. After thinking more, I have modified the patch to report
the NOTICE only when altering the retain_conflict_info for disabled
subscription as that aligns with the original intent for introducing the
NOTICE.

>
>
> 3)
> +
> +ALTER SUBSCRIPTION regress_testsub SET (slot_name = NONE); DROP
> +SUBSCRIPTION regress_testsub;
>
> Shall we enable regress_testsub as well before we drop it to catch unexpected
> notices/warnings if any?

I have added the ENABLE test in the tap-test in 0005, because we do not have a
valid connection for subscriptions created in subscription.sql, So enabling the
sub here could cause network error with random socket info, making the test
unstable.

>
>
> 4)
> postgres=# alter subscription sub2 disable;
> WARNING: commit timestamp and origin data required for detecting conflicts
> won't be retained
> HINT: Consider setting "track_commit_timestamp" to true.
>
> Do we need "track_commit_timestamp" related WARNING while disabling rci
> sub as well? I feel it should be there while enabling it alone.

Changed as suggested.

Here is the V35 patch set which includes the following changes:

0001:
Merged original V35-0001 and 0002.

0002:
* Addressed the comments above.
* The launcher will now attempt to acquire the conflict detection slot
initially, allowing us to determine its existence by checking if
MyReplicationSlot is NULL.
* Merged the top-up patch provided by Amit[1].

0003:
Rebased

0004:
Rebased

0005:
Added few tests to verify the WANRINGs and NOTCE.

0006:
Rebased

0007:
Rebased

Best Regards,
Hou zj