Re: proposal: add 'waiting for replication' to pg_stat_activity.state

On 3 December 2015 at 22:58, Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> wrote:

> On Thu, Dec 3, 2015 at 9:02 AM, Craig Ringer <craig(at)2ndquadrant(dot)com>
> wrote:
> >
> > On 3 December 2015 at 09:32, Peter Eisentraut <peter_e(at)gmx(dot)net> wrote:
> >>
> >> On 12/2/15 7:00 PM, Craig Ringer wrote:
> >> > I notice that you don't set the 'waiting' flag. 'waiting' is
> presently
> >> > documented as:
> >> >
> >> > <entry>True if this backend is currently waiting on a
> lock</entry>
> >> >
> >> > ... but I'm inclined to just widen its definition and set it here,
> since
> >> > we most certainly are waiting, and the column isn't named
> >> > 'waiting_on_a_lock'. It shouldn't upset various canned lock monitoring
> >> > queries people have since they generally do an inner join on pg_locks
> >> > anyway.
> >>
> >> I'm not so sure about that assumption.
> >
> >
> > Even if it's an outer join, the worst that'll happen is that they'll get
> entries with nulls in pg_locks. I don't think it's worth worrying about too
> much.
> >
>
> That can be one way of dealing with it and another is that we
> keep the current column as it is and add another view related
> wait stats, anyway we need something like that for other purposes
> like lwlock waits etc. Basically something on lines what we
> is being discussed in thread [1]
>
> [1] -
> http://www.postgresql.org/message-id/CAA4eK1+=5Ex8-5NRr3u94=_t2p65v0kcjZ5rXddVmkx=LwAJ+w@mail.gmail.com
>

Good point. I'm not sure this shouldn't set 'waiting' anyway, though.

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