[Pgpool-hackers] Major problem in find_primary_node()

[Gilles Darold]

Hi Tatsuo,

You have to use an existing ip address /host running postmaster like an
other slave. This host must not have a wal writer process.

Here I've attached an avi video that demonstatre the problem.

Regards,

Le 21/01/2011 03:02, Tatsuo Ishii a écrit :
>>>
>>> I know that the problem is difficult to reproduce, especially if every
>>> administration task is well done.
>>>
>>> What you can try is to reconstruct Node 0 with a valid recovery.conf,
>>> then status of the node will be 2, PgPool will know it's a standby
>>> server, node 1 is the master host and all it's ok at this point.
>>>
>>> Next step, just modify the recovery.conf of node 0 with a wrong ip
>>> address/hostname and then restart PostgreSQL. Node 0 in pool_status
>>> should stay in state 2 as secondary node which is still right.
>> Wrong IP/hostname, ok, I will try it.
> I tried but failed to succeed in reproducing your problem.
>
> Node 0 continues to print following errors(this is expected):
>
> 4491 2011-01-21 10:14:12 JST FATAL:  could not connect to the primary server: could not translate host name "foo" to address: Name or service not known
>
> And failed to connect to node 0 postmaster. So pgpool regards it as down.
>
> test=# show pool_nodes;
>  node_id | hostname | port | status | lb_weight 
> ---------+----------+------+--------+-----------
>  0       |          | 5433 | 3      | 0.500000
>  1       |          | 5434 | 2      | 0.500000
> (2 rows)
>
> Is there anything I have to do more to reproduce your problem?

-- 
Gilles Darold
http://dalibo.com - http://dalibo.org

-------------- next part --------------
A non-text attachment was scrubbed...
Name: bugdemo.avi.bz2
Type: application/x-bzip
Size: 1832526 bytes
Desc: not available
URL: <http://pgfoundry.org/pipermail/pgpool-hackers/attachments/20110121/4afc5e86/attachment-0001.bin>