Re: [HACKERS] strange parallel query behavior after OOM crashes

On Tue, Apr 11, 2017 at 2:36 AM, Robert Haas <robertmhaas@gmail.com> wrote:
> On Mon, Apr 10, 2017 at 2:32 PM, Neha Khatri <nehakhatri5@gmail.com> wrote:
>> On Tue, Apr 11, 2017 at 1:16 AM, Robert Haas <robertmhaas@gmail.com> wrote:
>>> 1. Forget BGW_NEVER_RESTART workers in
>>> ResetBackgroundWorkerCrashTimes() rather than leaving them around to
>>> be cleaned up after the conclusion of the restart, so that they go
>>> away before rather than after shared memory is reset.
>>
+1 for the solution.

>> Now with this, would it still be required to forget BGW_NEVER_RESTART
>> workers in maybe_start_bgworker():
>>
>>                 if (rw->rw_crashed_at != 0)
>> {
>> if (rw->rw_worker.bgw_restart_time == BGW_NEVER_RESTART)
>> {
>> ForgetBackgroundWorker(&iter);
>> continue;
>> }
>>                      ......
>>                 }
>
> Well, as noted before, not every background worker failure leads to a
> crash-and-restart; for example, a non-shmem-connected worker or one
> that exits with status 1 will set rw_crashed_at but will not cause a
> crash-and-restart cycle.   It's not 100% clear to me whether the code
> you're talking about can be reached in those cases, but I wouldn't bet
> against it.
I think that for above-mentioned background workers, we follow a
different path to call ForgetBackgroundWorker().
CleanupBackgroundWorker()
- ReportBackgroundWorkerExit()- ForgetBackgroundWorker()
But, I'm not sure for any other cases.

Should we include the assert statement as well?

-- 
Thanks & Regards,
Kuntal Ghosh
EnterpriseDB: http://www.enterprisedb.com