[Pgpool-general] Issue with pgPool, failures and disconnect

[Tatsuo Ishii]

> Folks,
> 
> OK, we've pinpointed the problem.  Let me step-by-step it:
> 
> 1) Serv1 is running pgpool, in single-server mode, connected to Serv2,which is 
> running PostgreSQL 8.0.2.
> 
> 2) App1 is connected to pgpool, and is running a series of SQL statements.
> 
> 3) Serv2 fails due to a hardware issue, and shuts down.
> 
> 4) pgpool does NOT drop the connection to JDBC.  It hangs on to it.
> 
> 5) JDBC hangs on to the connection, waiting for its last query to return.  For 
> hours.
> 
> The problem seems to be (4).   That is, if there is a server failure and there 
> is not a secondary server to fail over to, pgPool should terminate the 
> connecition.  Am I wrong?

If Server2 goes down and pgpool detects the failure, pgpool should
restart all child processes. Thus the connections between pgpool and
pgpool's clients will be forced to be terminated. 

In this case, you will see something like this in your pgpool log:

2005-05-26 08:59:04 LOG:   pid 2461: notice_backend_error: master: 1 fail over request from pid 2461
2005-05-26 08:59:04 LOG:   pid 2439: restarting pgpool
2005-05-26 08:59:04 LOG:   pid 2439: restarting pgpool done.

How was it?
--
Tatsuo Ishii