[pgpool-general: 6005] problems with pgpool-II and balanced Virtual IP

[Jaime Soler]

Hi,

we have next configuration:

Many applications connect to 3 pgpool-II with watchdog, 1 pgpool active and
2 pgpool standby.

We have stopped pgpool active and the applications have taken 40 minutes to
reconnect to pgpool.

Our log is that:

******************************************
old master pgpool (pgpool1) reboot
******************************************
2018-02-01 02:28:14: pid 14560: LOG:  remote node "pgpool1:30000 Linux
pgpool1" is shutting down
...
2018-02-01 02:28:14: pid 14560: LOG:  unassigning the remote node
"pgpool1:30000 Linux pgpool1" from watchdog cluster master
...

******************************************
At this time we can't connecting to the DB
******************************************
2018-02-01 02:28:14: pid 14557: ERROR:  unable to read data from DB node 0

************************************************
New pgpool (pgpool2) node becomes master
************************************************
2018-02-01 02:28:15: pid 14560: LOG:  watchdog node state changed from
[INITIALIZING] to [STANDING FOR MASTER]
2018-02-01 02:28:15: pid 14560: LOG:  watchdog node state changed from
[STANDING FOR MASTER] to [MASTER]
2018-02-01 02:28:15: pid 14560: LOG:  I am announcing my self as
master/coordinator watchdog node
2018-02-01 02:28:15: pid 14560: LOG:  I am the cluster leader node
2018-02-01 02:28:15: pid 14560: LOG:  setting the local node
"pgpool2:30000 Linux pgpool2" as watchdog cluster master
2018-02-01 02:28:15: pid 14560: LOG:  I am the cluster leader node but
we do not have enough nodes in cluster

***********************************************************
pgpool node (pgpool3) is adding to the standby list
***********************************************************
2018-02-01 02:28:16: pid 14560: LOG:  adding watchdog node
"pgpool3:20000 Linux pgpool3" to the standby list
2018-02-01 02:28:16: pid 14560: LOG:  quorum found

***********************************************
3 second there is a first connection to the DB
***********************************************
2018-02-01 02:28:17: pid 14560: LOG:  new IPC connection received

***********************************************************
pgpool node (pgpool1) is adding to the standby list
***********************************************************
2018-02-01 02:35:17: pid 14560: LOG:  new node joined the cluster
hostname:"pgpool1" port:30000 pgpool_port:30000


I have any questions:

¿how it work arping in that case? ¿arping change arptables?

One of the problem that we have is that VirtualIP not has change from
pgpool1 (stop) to pgpool2 (active), then we can connecting from psql in the
same network but we cannot connecting  from applications. The applications
are in other subnet.

The problem are not the permission because to the 40 minutes the
applications can connecting and before stop pgpool1 too.

Thanks.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.sraoss.jp/pipermail/pgpool-general/attachments/20180320/0acb52f0/attachment.html>