[pgpool-general: 6404] Re: How to measure or detect run out of connection from pgpool perspective ?

Tatsuo Ishii ishii at sraoss.co.jp
Sun Feb 3 08:11:52 JST 2019


> Hi,
> 
> I have setup a two server cluster replicate using streaming replication and
> watchdog is handling vip. Here are the general connections settings:
> pgpool.conf:
> num_init_children = 128
> max_pool = 4
> listen_backlog_multiplier = 2
> 
> postgresql.conf
> max_connections = 512
> superuser_reserved_connections = 3
> 
> How to measure or detect run out of connection from pgpool perspective ?

Currently there's no way to measure run out of connection to frontend.

> is it help counting  last column pool_connected equal 1 of command
> pcp_proc_info

Yes. Count number of lines emitted by pcp_proc_info (without -a). Let
the number be n.  If (n / (number of live backend)) =
num_init_children, Pgpool-II runs out connection.

> -a  ? if pgpool run out of connection will give any hint or log message on
> pgpool's log ?

No.

BTW, upcoming Pgpool-II 4.1 will have a capability if connection is
run out, no additional connection is accepted (similar behavior to
PostgreSQL).

See "reserved_connections" parameter described in:
http://tatsuo-ishii.github.io/pgpool-II/current/runtime-config-connection.html#RUNTIME-CONFIG-CONNECTION-SETTINGS

Best regards,
--
Tatsuo Ishii
SRA OSS, Inc. Japan
English: http://www.sraoss.co.jp/index_en.php
Japanese:http://www.sraoss.co.jp


More information about the pgpool-general mailing list