[Pgpool-general] pgpool-II replication stopped working after couple of weeks

Asaf asaf at lingnu.com
Tue Oct 20 14:04:09 UTC 2009


Tatsuo Ishii wrote:

>> i'm trying to run pgpool as replication solution, it worked fine for 
>> couple of weeks,
>>
>> then a 'create table' didn't work on the mirror, after stopping and 
>> starting pgpool-II,
>>
>> it worked again.
>>
>> how can i debug this ?
>>     
>
> When you are saying "create table didn't work", what do you see in
> pgpoo and PostgreSQL log? That's the starting point.
>
>   
>> what does 'detect_error: kind: Z' means ?
>>     
>
> It's normal.
> --
> Tatsuo Ishii
> SRA OSS, Inc. Japan
>
>   

I continue trying, I didn't have detailed enough logs in the first time, 
now i do,
the replication gets out of sync every couple of days, I'm investigating 
the logs now
to understand whats gone wrong, and thats what i see :

at some point in time, the log of the secondary backend is finished, no 
more activity,
although postgres continue to run as normal (suppose to serve only for 
pgpool-II)
I look in pgpool log at same point in time, seeing nothing unusual,
(but users get  errors saying : ERROR:  pgpool detected difference of 
the number of inserted, updated or deleted tuples. Possible last query 
was:...)

looks like at some point pgpool stops sending queries to the secondary 
backend, but, again,
nothing in pgpool logs, even NUM_BACKENDS says '2', (although 
occasionally it also says '1')

how can I debug this ?
10x

-- 
Asaf Ohaion
Lingnu Open Source Consulting Ltd
09-7655871 (204), 0502-565023
http://www.lingnu.com




More information about the Pgpool-general mailing list