[Postgres-xl-developers] Possible regression with recent GTM patches in XL9_5_STABLE

Krzysztof Nienartowicz Krzysztof.Nienartowicz at unige.ch
Sun Oct 22 04:39:36 PDT 2017


Sure,
PM sent.
Cheers

> On 22 Oct 2017, at 13:35, Tomas Vondra <tomas.vondra at 2ndquadrant.com> wrote:
> 
> 
> 
> On 10/21/2017 10:09 PM, Krzysztof Nienartowicz wrote:
>> Hello Tomas,
>> 
>> We have experienced failed processing due to shared queue problems and
>> segfaults under load with the latest two patches from 95 branch on
>> queries that had run before without problems.
>> 
>> * Allocate thread-local snapshot array statically
>> * Remember queryId for queries executed using FQS
>> 
>> We did not have time to analyze in detail but reverting patches fixes
>> the problem (no idea which one could be a culprit..).
>> Cores are available just in case.
>> 
> 
> Damn. I don't quite see how the queryId could cause any failures, so I
> assume it has to be the other patch.
> 
> Can you please share the backtraces from the core files?
> 
> 
> regards
> 
> -- 
> Tomas Vondra                  http://www.2ndQuadrant.com
> PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services



More information about the Postgres-xl-developers mailing list