Julien Rouhaud writes:
> On Fri, Apr 19, 2019 at 7:17 PM Tom Lane wrote:
>> My thought was that we'd like this to work without requiring any new
>> server-side facilities, so that pg_dump could use it against any server
>> version that supports parallel dump.
> Couldn't we use LOCKTAG_USERLOCK f
On Fri, Apr 19, 2019 at 7:17 PM Tom Lane wrote:
>
> Robert Haas writes:
> > On Wed, Apr 17, 2019 at 11:34 AM Tom Lane wrote:
> >> ... If there are user applications
> >> running that also use advisory locks, there could be unwanted
> >> interference. One easy improvement is to use pg_try_adviso
Robert Haas writes:
> On Wed, Apr 17, 2019 at 11:34 AM Tom Lane wrote:
>> ... If there are user applications
>> running that also use advisory locks, there could be unwanted
>> interference. One easy improvement is to use pg_try_advisory_lock(k) in
>> step 2, and just choose a different k if the
On Wed, Apr 17, 2019 at 11:34 AM Tom Lane wrote:
> While thinking about that, it occurred to me that we could close the
> gap if the server somehow understood that the master was waiting for
> the worker. And it's actually not that hard to make that happen:
> we could use advisory locks. Conside
