[Pgpool-general] Reloading the server while under load and recovering the degenerated node

[Jaume Sabater]

On Tue, Feb 24, 2009 at 4:18 PM, Tatsuo Ishii <ishii at sraoss.co.jp> wrote:

>> Regarding the "some times client connections are not dropped and the
>> online recovery process fails at second stage", any hints or ideas?
>
> recovery_timeout must be much longer than
> client_idle_limit_in_recovery.
>
> Are you sure with this?

At present time these are the values of those directives:

recovery_timeout = 90
client_idle_limit_in_recovery = 30

Do you think these are good enough?

Also, what would you suggest to help debug this problem? Which sort of
tests would you advice me to do or try?

Thanks.

-- 
Jaume Sabater
http://linuxsilo.net/

"Ubi sapientas ibi libertas"