Obsolete socket files if not shut down cleanly

[tyldum]

If the container is not stopped cleanly there are leftover sockets in /var/run/postgresql which will cause the container to refuse to run. Works if you manually delete /var/run/postgresql/.s.PGSQL.* or delete the container.

Might be better handled upstream?

[sqilsoft]

Some problem:

STARTING SSH (if required)... TUNING UP SSH CLIENT... STARTING SSH SERVER... TURNING PGPOOL... Opening access from all hosts by md5 in /usr/local/etc/pool_hba.conf Adding user pcp_user for PCP Creating a ~/.pcppass file for pcp_user Adding users for md5 auth

Adding user monkey_user Adding check user 'monkey_user' for md5 auth Adding user 'monkey_user' as check user Adding user 'monkey_user' as health-check user Adding backends Waiting for backend 0 to start pgpool (WAIT_BACKEND_TIMEOUT=60) 2018/03/15 11:31:16 Waiting for host: tcp://pgmaster:5432 2018/03/15 11:31:16 Connected to tcp://pgmaster:5432 Adding backend 0 Waiting for backend 1 to start pgpool (WAIT_BACKEND_TIMEOUT=60) 2018/03/15 11:31:16 Waiting for host: tcp://pgslave1:5432 2018/03/15 11:31:32 Connected to tcp://pgslave1:5432 Adding backend 1 Waiting for backend 3 to start pgpool (WAIT_BACKEND_TIMEOUT=60) 2018/03/15 11:31:32 Waiting for host: tcp://pgslave3:5432 2018/03/15 11:31:32 Connected to tcp://pgslave3:5432 Adding backend 3 Waiting for backend 2 to start pgpool (WAIT_BACKEND_TIMEOUT=60) 2018/03/15 11:31:32 Waiting for host: tcp://pgslave2:5432 2018/03/15 11:31:51 Connected to tcp://pgslave2:5432 Adding backend 2 Checking if we have enough backends to start Will start pgpool REQUIRE_MIN_BACKENDS=3, BACKENDS_COUNT=4 Configuring /usr/local/etc/pgpool.conf Adding config 'num_init_children' with value '250' Adding config 'max_pool' with value '4' STARTING PGPOOL... 2018-03-15 11:31:51: pid 73: FATAL: failed to bind a socket: "/var/run/postgresql/.s.PGSQL.5432" 2018-03-15 11:31:51: pid 73: DETAIL: bind socket failed with error: "Address already in use"