[pgpool-general: 2252] Re: pgpool 3.3.1 replicate setup with 3 nodes

[Yugo Nagata]

Hi,

Are you setting recovery_2nd_stage_command properly?
(http://www.pgpool.net/docs/latest/pgpool-en.html#online-recovery
 http://www.pgpool.net/docs/latest/pgpool-en.html#RECOVERY_2ND_STAGE_COMMAND)

If this is wrong, recovery might fails and backends might get inconsistent.
For analysis, could you please provide your pgpool.conf and all scripts? 

On Wed, 6 Nov 2013 06:08:05 -0800 (PST)
Videanu Adrian <videanuadrian at yahoo.com> wrote:

> Hi all,
> 
> I have a pgpool 3.3.1 with 3 postgresql 9.2 backends.
>  The pgpool is configured as : [ mode ]
>     Replication Mode
>  / Load Balance Mode / Watchdog On
> 
> I have a problem when I use online recovery for one of the nodes.
> 
> Scenario: 
> 1. The 3rd node failed.
> 2. Pgpool detects the failure and present the node as down. (everything is ok)
> 3. I press recovery from pgpadmin and the recovery script starts.
> 4. The 3rd node is back online, but the sequence for this node and the 1st node are incremented by 1 regarding the 2nd node.
> Because of this nothing works because i get a kind mismatch between nodes.
> 
> Does replication setup works for more than 2 nodes and if yes how should i perform online recovery.
> 
> PS. I use PITR.
> 
> 
>  
> Regards,
> Adrian Videanu


-- 
Yugo Nagata <nagata at sraoss.co.jp>