[Pgpool-general] Replication health check failed on 2nd node

[Tatsuo Ishii]

>  On 09/28/10 19:19, Tatsuo Ishii wrote:
>>> When I started the pgpool2 (version 3.0 on CentOS, with user
>>> 'sd3ops1', not 'postgres') on a host different from the two db hosts,
>>> I noticed that the log always failed on health_check on node 1 db with
>>> the following message (without any client connection yet):
>>>
>>> health_check: 0 th DB node status:1
>>> health_check: 1 th DB node status:3
>>>
>>> I tried to switch the db hosts and still got the same results.  I also
>>> tried manually without problem to connect to each db.  Any solution
>>> and/or suggestion on how to resolve the problem ?
>> Maybe health_check_user is not appropreate or pg_hba.conf on the 1th
>> DB
>> does not accept connection from pgpool-II.
> This may not be related, since as I mentioned when I switched the db
> hosts
> in the pgpool.conf, I still got the same error on the 1th DB.
>> What does it show if you run psql on the host where pgpool-II is
>> running:
>>
>> psql -U (the username you specfied for health_check_user) -h (the host
>> you have problem with) template1
> I tried it with both DBs and they all ask for password.
> 
> I can still connect to the pgpool to do the query, but since 1th DB is
> offline, I think it is from the 0th DB.
> 
> Any other suggestions to try to find out the problem ?

I suspect pgpool thinks that 1th DB is down by looking at pgpool_status file.
Can you stop and start pgpool with -D option? Which ignores pgpool_status file.
--
Tatsuo Ishii
SRA OSS, Inc. Japan
English: http://www.sraoss.co.jp/index_en.php
Japanese: http://www.sraoss.co.jp