Re: Query running for very long time (server hanged) with parallel append

On 6 February 2018 at 10:11, Kyotaro HORIGUCHI
<horiguchi(dot)kyotaro(at)lab(dot)ntt(dot)co(dot)jp> wrote:
>> At Mon, 5 Feb 2018 15:29:27 +0530, Amit Khandekar <amitdkhan(dot)pg(at)gmail(dot)com> wrote in
>> > Attached is a patch that fixes this issue on the above lines.
>>
>> The patch adds two new variables and always sets them but warp
>> around can happen at most once per call so I think it is enough
>> to arrange to stop at the wrap around time. Addition to that the
>> patch is forgetting the case of no partial plan. The loop can
>> overrun on pa_finished in the case.
>
> Sorry, the patch works fine. But still the new variables don't
> seem needed.

True, I could have set initially as_whichplan to pa_next_plan, and
used as_whichplan instead of initial_plan. And, I could have used the
condition initial_plan >= append->first_partial_plan instead of
should_wrap_around. The reason I used these two variables is because I
thought the logic would be more reader-friendly. (Also,
should_wrap_around uses static values so using this variable avoids
determining the condition (initial_plan >= append->first_partial_plan)
at each iteration).

--
Thanks,
-Amit Khandekar
EnterpriseDB Corporation
The Postgres Database Company