On Wed, Jan 13, 2021 at 10:51 PM John Naylor
<john.naylor@enterprisedb.com> wrote:
>
>
> On Tue, Jan 12, 2021 at 9:37 PM Fujii Masao <masao.fujii@gmail.com> wrote:
> >
> > Thanks for the patch! I think this change makes sense.
> >
> > - (errmsg("terminating walsender process
> > due to replication timeout")));
> > + (errmsg("terminating walsender process
> > due to WAL sender timeout")));
> >
> > Isn't it a bit strange to include different expressions "walsender" and
> > "WAL sender" for the same thing in one message?
>
> It is strange, now that I think about it. My thinking was that the former wording of "replication timeout" was a less
literalreference to the replication_timeout parameter, so I did the same for wal_sender_timeout. A quick look shows we
arenot consistent in the documentation as far as walsender vs. WAL sender. For the purpose of the patch I agree it
shouldbe consistent within a single message. Maybe the parameter should be spelled exactly as is, with underscores?
I'm ok with that. But there seems no other timeout messages using
the parameter name.
src/backend/replication/logical/worker.c: (errmsg("terminating logical
replication worker due to timeout")));
src/backend/replication/walreceiver.c: (errmsg("terminating
walreceiver due to timeout")));
src/backend/replication/walsender.c: (errmsg("terminating walsender
process due to replication timeout")));
src/backend/tcop/postgres.c: errmsg("canceling authentication due to
timeout")));
src/backend/tcop/postgres.c: errmsg("canceling statement due to lock
timeout")));
src/backend/tcop/postgres.c: errmsg("canceling statement due to
statement timeout")));
src/backend/tcop/postgres.c: errmsg("terminating connection due to
idle-in-transaction timeout")));
src/backend/tcop/postgres.c: errmsg("terminating connection due to
idle-session timeout")));
src/backend/utils/misc/timeout.c: errmsg("cannot add more timeout reasons")));
> I'll take a broader look and send an updated patch.
Thanks!
Regards,
--
Fujii Masao