Re: shm_mq_wait_internal gets stuck forever on fast shutdown

On Mon, Aug 21, 2017 at 10:07 AM, Craig Ringer <craig(at)2ndquadrant(dot)com> wrote:
> Makes sense, and I'm not especially concerned. If the expected solution to
> such usage is to use non-blocking calls, that's fine with me.
>
> I partly wanted to put this out there to help the next person looking into
> it. Or myself, when I've forgotten and go looking again ;) . But also, to
> ensure that this was in fact fully expected behaviour not an oversight re
> applying shm_mq to non-bgworker endpoints.

Yep, it's expected. It's possible I should have designed it
differently, so if someone does feel concerned at some point we can
certainly debate how to change things, but what you're describing
matches my expectations and it seems OK to me, pretty much.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company