On Wed, Apr 13, 2016 at 4:54 PM, Michael Paquier
<michael.paquier@gmail.com> wrote:
> On Fri, Apr 8, 2016 at 4:49 PM, Fujii Masao <fujii@postgresql.org> wrote:
>> Add regression tests for multiple synchronous standbys.
>>
>> Authors: Suraj Kharage, Michael Paquier, Masahiko Sawada, refactored by me
>> Reviewed-By: Kyotaro Horiguchi
>
> Well, we are not quite there yet:
> http://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=hamster&dt=2016-04-12%2016%3A00%3A06
>
> # Running: pg_ctl -D
> /home/buildfarm/data/buildroot/HEAD/pgsql.build/src/test/recovery/tmp_check/data_master_Qmuz/pgdata
> reload
> server signaled
> not ok 2 - asterisk in synchronous_standby_names
>
> # Failed test 'asterisk in synchronous_standby_names'
> # at t/007_sync_rep.pl line 26.
> # got: 'standby1|1|sync
> # standby2|1|potential
> # standby3|0|async'
> # expected: 'standby1|1|sync
> # standby2|1|potential
> # standby3|1|potential'
This seems to be a timing issue.
There can be small window after SIGHUP is sent before walsender updates
its priority based on new s_s_names. If pg_stat_replication is checked
before that update, it displays unexpected output. Probably we need to
sleep a few second after pg_ctl reload before pg_stat_replication.
Regards,
--
Fujii Masao
