[pgpool-general: 3903] Re: pgpool doesn't see the new master on failover

[Maksym Filenko]

Hey Muhammad,

thank you for your prompt reply!

Any ideas how could it happen? I've got following in my pgpool.conf:

backend_hostname0 = 'db1.tx.dev.mydomain'
backend_port0 = 5432
backend_weight0 = 1
backend_data_directory0 = '/var/lib/postgresql/9.4/main'
backend_flag0 = 'ALLOW_TO_FAILOVER'

backend_hostname1 = 'db2.tx.dev.mydomain'
backend_port1 = 5432
backend_weight1 = 1
backend_data_directory1 = '/var/lib/postgresql/9.4/main'
backend_flag1 = 'ALLOW_TO_FAILOVER'

fail_over_on_backend_error = on

Name: db1.tx.dev.mydomain
Address: 10.200.12.33

Name: db2.tx.dev.mydomain
Address: 10.200.12.34

Could you also please point a line from logs which leads you to this
conclusion?

Thanks a lot!

Regards,
Max

On Thu, Jul 30, 2015 at 12:36 PM, Muhammad Usama <m.usama at gmail.com> wrote:

> Hi Max
>
> Thanks for sharing the logs. From the log it seems like both backend0 and
> backend1 are pointing to the same PostgreSQL server, so when that PG server
> goes down(which is actually both master and slave as per the
> configuration). pgpool-II is left with no valid PostgreSQL backend, As a
> result no new master node gets selected and eventually the failover.sh gets
> empty values for new master hostname (%H) and database cluster path (%R)
> arguments.
>
> Thanks and regards
> Muhammad Usama
>
> On Wed, Jul 29, 2015 at 9:28 PM, Maksym Filenko <
> maksym.filenko at kreditech.com> wrote:
>
>> Hey Muhammad,
>>
>> thank you for your reply!
>>
>> Please, find gzipped logs attached below. pgpool-1 is for master,
>> pgpool-2 is for standby. Master Postgres node was shutted down at 18:01:01.
>> You can see a failover.sh call with insufficient parameters. It should be
>> something like '/srv/pgsql/bin/failover.sh 0 db2.tx.dev.mydomain' (new
>> master hostname).
>>
>> Again, thank you!
>>
>> On Wed, Jul 29, 2015 at 10:53 AM, Muhammad Usama <m.usama at gmail.com>
>> wrote:
>>
>>> Hi
>>>
>>> Can you please share the pgpool-II log with debug enabled.
>>>
>>> Thanks
>>> Best regards
>>> Muhammad Usama
>>>
>>>
>>> On Mon, Jul 27, 2015 at 8:18 PM, Maksym Filenko <
>>> maksym.filenko at kreditech.com> wrote:
>>>
>>>> Hey everyone,
>>>>
>>>> I've set up exactly this configuration:
>>>> http://www.pgpool.net/pgpool-web/contrib_docs/watchdog_master_slave_3.3/en.html
>>>>
>>>> I have two instances of pgpool-II version 3.3.4 (tokakiboshi) in front
>>>> of two PostgreSQL 9.4.3.
>>>>
>>>> I have following in my pgpool.conf:
>>>>
>>>> failover_command = '/var/lib/postgresql/9.4/main/failover.sh %d %P %H
>>>> %R'
>>>>
>>>> When master goes down and failover occurs, pgpool passes nothing to
>>>> failover.sh as the new master hostname (%H) and database cluster path (%R):
>>>>
>>>> $ grep 'failover' /var/log/messages
>>>> failover_handler: no valid DB node found
>>>> execute command: /var/lib/postgresql/9.4/main/failover.sh 0 0
>>>> failover: set new primary node: -1
>>>>
>>>> If the slave fails, everything works as expected:
>>>>
>>>> $ grep 'failover' /var/log/messages
>>>> execute command: /var/lib/postgresql/9.4/main/failover.sh 1 0
>>>> db1.example.com /var/lib/postgresql/9.4/main
>>>> failover: set new primary node: 0
>>>> failover: set new master node: 0
>>>>
>>>> Looks like pgpool doesn't see the slave to promote it.
>>>>
>>>> Any clue would be extremely appreciated!
>>>>
>>>> Best regards,
>>>> Max
>>>>
>>>> _______________________________________________
>>>> pgpool-general mailing list
>>>> pgpool-general at pgpool.net
>>>> http://www.pgpool.net/mailman/listinfo/pgpool-general
>>>>
>>>>
>>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.sraoss.jp/pipermail/pgpool-general/attachments/20150730/800f7a2c/attachment.html>