Re: Fix instability in subscription regression test

On Thu, Jun 23, 2022 at 8:28 PM houzj(dot)fnst(at)fujitsu(dot)com
<houzj(dot)fnst(at)fujitsu(dot)com> wrote:
>
> Hi,
>
> I noticed BF member wrasse failed in 028_row_filter.pl.
>
> # Failed test 'check publish_via_partition_root behavior'
> # at t/028_row_filter.pl line 669.
> # got: ''
> # expected: '1|100
> # ...
>
> Log:
> 2022-06-23 11:27:42.387 CEST [20589:3] 028_row_filter.pl LOG: statement: ALTER SUBSCRIPTION tap_sub REFRESH PUBLICATION WITH (copy_data = true)
> 2022-06-23 11:27:42.470 CEST [20589:4] 028_row_filter.pl LOG: disconnection: session time: 0:00:00.098 user=nm database=postgres host=[local]
> 2022-06-23 11:27:42.611 CEST [20593:1] LOG: logical replication table synchronization worker for subscription "tap_sub", table "tab_rowfilter_partitioned" has started
> ...
> 2022-06-23 11:27:43.197 CEST [20610:3] 028_row_filter.pl LOG: statement: SELECT a, b FROM tab_rowfilter_partitioned ORDER BY 1, 2
> ...
> 2022-06-23 11:27:43.689 CEST [20593:2] LOG: logical replication table synchronization worker for subscription "tap_sub", table "tab_rowfilter_partitioned" has finished
>
> From the Log, I can see it query the target table before the table sync is
> over. So, I think the reason is that we didn't wait for table sync to
> finish after refreshing the publication. Sorry for not catching that
> ealier. Here is a patch to fix it.

+1

The patch looks good to me.

Regards,

--
Masahiko Sawada
EDB: https://www.enterprisedb.com/