[pgpool-general: 6769] Re: Watchdog does not get consensus for failover

[shacky]

In every tests I done I waited much more than 40 seconds (at least 5
minutes).
The strange thing is that failover was working until 1 week (when I wrote
my first message in this thread). I made 2 successfully tests, but after
the third tests the failover did not work anymore. I did not change
anything in the configuration file.
Now I'm just thinking to restart the whole PgPool from scratch, but I wish
to solve this problem before going in production.

Il giorno lun 4 nov 2019 alle ore 14:52 Muhammad Usama <m.usama at gmail.com>
ha scritto:

>
>
> On Sat, Nov 2, 2019 at 2:19 PM shacky <shacky83 at gmail.com> wrote:
>
>> Thanks Muhammad!
>>
>>
>>> Can you share the pgpool.conf files for all nodes. My suspicion is that
>>> the health-check timeout values
>>> are somewhat on the higher side, So that's why the  health-check might
>>> have missed the unavailability of
>>> PG server.
>>>
>>
>> Here you can download the pgpool.conf file running on every PgPool node
>> (I removed sensitive data): https://we.tl/t-5Zj7o8BK8x
>>
>
> All pgpool.conf files are using 40-second health check period, That means
> a pgpool-II node can take up to 40 seconds to detect a backend node
> failure. In your test how long did you waited for the pgpool to make a
> failover consensus?
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.sraoss.jp/pipermail/pgpool-general/attachments/20191104/186380d1/attachment.html>