Re: group locking: incomplete patch, just for discussion

On Sat, Nov 1, 2014 at 5:06 PM, Andres Freund <andres(at)2ndquadrant(dot)com> wrote:
> On 2014-11-01 17:00:59 -0400, Robert Haas wrote:
>> On Sat, Nov 1, 2014 at 1:55 PM, Andres Freund <andres(at)2ndquadrant(dot)com> wrote:
>> > I doubt it. There's a whole bunch of problems that just exist by virtue
>> > of having a group leader. What if the leader dies but the worker backend
>> > isn't in a section of code that does a CHECK_FOR_INTERRUPTS()?
>>
>> In between all of the heat about whether I'm writing the correct
>> patch, or whether I'm overall making progress fast enough, somebody
>> could, I don't know, read the patch.
>
> I plan to, but that takes a fair amount of time. And there's at least
> one patch of yours ahead in the queue... And I do think that most of the
> concerns are more general than a specific implementation.

Fair enough. Well, the way I solved that specific problem is to allow
the group leader to terminate but not to return its PGPROC to the free
list until the last group member exits.

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