[pgpool-general: 5566] Re: pgpool and erlang - request hangs

Pavel balroga3 at yandex.ru
Wed Jun 7 17:21:26 JST 2017


Hello, Tatsuo.

We've found no additional problems so far, thank you very very very much!

You wrote 5 июня 2017 г., 11:46:49:

>> Hello, Tatsuo.
>> 
>> Thank  you  very much, test connections works normally!!!
>> We'll run more tests today and I'll report the results.

> I'm looking forward to hearing the results.

>> What   about  the  changes  you  proposed in patch in "pgpool-general:
>> 5506",  should  I keep them as well? Right now I'm running pgpool with
>> both patches applied.

> Yes, the patch is also necessary.

> Best regards,
> --
> Tatsuo Ishii
> SRA OSS, Inc. Japan
> English: http://www.sraoss.co.jp/index_en.php
> Japanese:http://www.sraoss.co.jp

>> You wrote 5 июня 2017 г., 8:50:13:
>> 
>>> Ok, I think I found a another possible cause.  After the client sends
>>> a flush message, it read the socket until Pgpool-II returns a
>>> response. Unfortunately, Pgpool-II keeps on waiting response from
>>> backend, because after "Row description" message returned from backend
>>> because "query in progress" in Pgpool-II is on. Attached patch resets
>>> the flag upon receiving "Row description" message. I hope this solves
>>> your problem.
>> 
>>> Best regards,
>>> --
>>> Tatsuo Ishii
>>> SRA OSS, Inc. Japan
>>> English: http://www.sraoss.co.jp/index_en.php
>>> Japanese:http://www.sraoss.co.jp
>> 
>>>>> Hello Tatsuo.
>>>>> 
>>>>> Did  you have an opportunity to look into our problem? Is there anything I
>>>>> can do to help you nail this bug?
>>>> 
>>>> I was a little bit busy past 1 week due to attending PGCon in
>>>> Ottawa. I will look into this. BTW, strange thing is, I was succeeded
>>>> in reproducing your problem by using my own test tool and confirmed
>>>> the fix works. Probably there are more causes which I am missing in
>>>> your case.
>>>>>                                         (errmsg("Entered patched location"),
>>>>>                                              errdetail("message type was %c(%02x)", fkind, fkind)));
>>>>>                                         pool_set_doing_extended_query_message();
>>>>>                                 }
>>>>>                                 status = SimpleForwardToBackend(fkind, frontend, backend, len, contents);
>>>>>                                 break;
>>>>>                         }
>>>>> 
>>>>> Indeed, debug log contains records
>>>>>>DEBUG:  Entered patched location
>>>>>>DETAIL:  message type was H(48)
>>>>>>LOCATION:  pool_proto_modules.c:2484
>>>>> 
>>>>> To sum up:
>>>>> replication_mode = off
>>>>> load_balance_mode = on
>>>>> master_slave_mode = on
>>>>> master_slave_sub_mode = 'stream'
>>>>> memory_cache_enabled = off
>>>>> 
>>>>> If master_slave_mode is off, then connection from erlang library does not hang.
>>>> 
>>>> Thank you for additional information.
>>>> 
>>>> Best regards,
>>>> --
>>>> Tatsuo Ishii
>>>> SRA OSS, Inc. Japan
>>>> English: http://www.sraoss.co.jp/index_en.php
>>>> Japanese:http://www.sraoss.co.jp
>>>> _______________________________________________
>>>> pgpool-general mailing list
>>>> pgpool-general at pgpool.net
>>>> http://www.pgpool.net/mailman/listinfo/pgpool-general
>> 
>> 
>> 
>> -- 
>> Best regards,
>>  Pavel                          mailto:balroga3 at yandex.ru
>> 



-- 
Best regards,
 Pavel                          mailto:balroga3 at yandex.ru



More information about the pgpool-general mailing list