[pgpool-general: 6138] Re: Disable failover for a single pgpool/postgres node configuration

Pierre Timmermans ptim007 at yahoo.com
Sat Jun 23 19:03:19 JST 2018

There are a lot of parameters and it is indeed difficult to choose the right combination :-), so I am not sure neither.

>From my experiments, I see a difference if you set the flag DISALLOW_TO_FAILOVER versus ALLOW_TO_FAILOVER. 

with ALLOW_TO_FAILOVER, when the postgres database comes down pgpool will be detach it from the pool. But when the database is started again it won't be re-attached so you will have to do it manually.

with DISALLOW_TO_FAILOVER you don't have this problem, in fact can not use the pcp_detach_node functionality
So I believe setting DISALLOW_TO_FAILOVER is better in a single node set-up. I would also set fail_over_on_backend_error to no and health_check_period = 0 and failover_command = '' (empty string) and load_balance_mode = no
In a multi-node set-up, i found that having DISALLOW_TO_FAILOVER is not OK because then you cannot use pcp_detach (in case for example you would manually want to exclude one standby from the pool)



    On Friday, June 22, 2018, 5:53:19 PM GMT+2, Alexander Dorogensky <amazinglifetime at gmail.com> wrote:  
I'm running pgpool 3.6.10 with postgres 9.6.8 in a single node configuration. Currently fail_over_on_backend_error = on, so I believe setting it to off will disable it?
Do I also have to set backend_flag0 to DISALLOW_FAILOVER?
pgpool-general mailing list
pgpool-general at pgpool.net
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.sraoss.jp/pipermail/pgpool-general/attachments/20180623/87e49254/attachment.html>

More information about the pgpool-general mailing list