Re: bgw_type (was Re: Why does logical replication launcher set application_name?)

On 8/31/17 23:22, Michael Paquier wrote:
> On Fri, Sep 1, 2017 at 4:49 AM, Peter Eisentraut
> <peter(dot)eisentraut(at)2ndquadrant(dot)com> wrote:
>> On 5/30/17 23:10, Peter Eisentraut wrote:
>>> Here is a proposed solution that splits bgw_name into bgw_type and
>>> bgw_name_extra. bgw_type shows up in pg_stat_activity.backend_type.
>>> Uses of application_name are removed, because they are no longer
>>> necessary to identity the process type.
>>
>> Updated patch incorporating the feedback. I have kept bgw_name as it
>> was and just added bgw_type completely independently.
>
> - errmsg("terminating background worker \"%s\" due to
> administrator command",
> - MyBgworkerEntry->bgw_name)));
> + errmsg("terminating %s due to administrator command",
> + MyBgworkerEntry->bgw_type)));
> "terminating background worker %s of type %s due to administrator
> command", bgw_name, bgw_type?

OK.

>> One open question is how to treat a missing (empty) bgw_type. I
>> currently fill in bgw_name as a fallback. We could also treat it as an
>> error or a warning as a transition measure.
>
> Hm. Why not reporting an empty type string as NULL at SQL level and
> just let it empty them? I tend to like more interfaces that report
> exactly what is exactly registered at memory-level, because that's
> easier to explain to users and in the documentation, as well as easier
> to interpret and easier for module developers.

The problem here is that we refer to bgw_type in a bunch of places now,
and adding a suitable fallback in all of these places would be a lot of
code and it would create a regression in behavior. In practice, I think
that would be a lot of trouble for no gain.

--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services