Re: [HACKERS] exposing wait events for non-backends (was: Trackingwait event for latches)

On 13 December 2016 at 01:45, Kevin Grittner <kgrittn@gmail.com> wrote:
> On Mon, Dec 12, 2016 at 10:33 AM, Andres Freund <andres@anarazel.de> wrote:
>> On 2016-12-12 13:26:32 -0300, Alvaro Herrera wrote:
>>> Robert Haas wrote:
>
>>>> 1. Show all processes that have a PGPROC in pg_stat_activity,
>
>>>> 2. Add a second view, say pg_stat_system_activity,
>
>>> I vote 1.
>>
>> +1
>
> +1

I've long wanted the ability to see auxillary process state in
pg_stat_activity, so +1.

Right now pg_stat_replication is a join over pg_stat_get_activity()
and pg_stat_get_wal_senders() filtered for walsenders, and
pg_stat_activity is a view over pg_stat_get_activity() filtered for
processes with a user id. I'd really like to see walsenders in
pg_stat_activity now that logical decoding makes them more than dumb
I/O channels, as well as other auxillary processes.

We should probably expose a proc_type or something, with types:

* client_backend
* bgworker
* walsender
* autovacuum
* checkpointer
* bgwriter

for simpler filtering.

I don't think existing user code is likely to get upset by more
processes appearing in pg_stat_activity, and it'll be very handy.

-- Craig Ringer                   http://www.2ndQuadrant.com/PostgreSQL Development, 24x7 Support, Training & Services