Slave don't connect to master

[jmcollin78]

Hello,

First I would like to thank you for this very usefull Docker clusterisation of Postgresql which is a mess.

I've got a strange behaviour I would like to expose when trying to start containers once by once (and not with Docker compose):

1. I start the first pg Container as master with this command:

docker run --rm --name Pg1 --env INITIAL_NODE_TYPE=master --env NODE_ID=1 --env NODE_NAME=node1 --env CLUSTER_NODE_NETWORK_NAME=pgmaster --env POSTGRES_PASSWORD=monkey_pass --env POSTGRES_USER=monkey_user --env POSTGRES_DB=monkey_db --env
 CONFIGS="listen_addresses:'*',logging_collector:on" --env CLUSTER_NAME=pg_cluster --env REPLICATION_DB=replication_db --env REPLICATION_USER=replication_user --env REPLICATION_PASSWORD=replication_pass --env FORCE_CLEAN=1 -p 5432:5432 -v
 /data/pg1:/var/lib/postgresql/data paunin/postgresql-cluster-pgsql

The PG is active and all seams to be OK.

2. I try to start the 2nd pg Container as slave with this command:

docker run --rm --name Pg2 --env INITIAL_NODE_TYPE=standby --env NODE_ID=2 --env NODE_NAME=node2 --env CLUSTER_NODE_NETWORK_NAME=pgstandby --env CONFIGS="listen_addresses:'*',logging_collector:'on'" --env CLUSTER_NAME=pg_cluster --env REP
LICATION_UPSTREAM_NODE_ID=1 --env FORCE_CLEAN=1 --env REPLICATION_PRIMARY_HOST=localhost --env REPLICATION_PRIMARY_PORT=5432 -p 5433:5432 -v /data/pg2:/var/lib/postgresql/data paunin/postgresql-cluster-pgsql

but the PG embedded in this container never start and never connect to master. Here are the logs:

For node with initial type standby you have to setup REPLICATION_UPSTREAM_NODE_ID
>>> Will wait db replication_db on pgstandby:5432, will try 50 times with delay 5 seconds
>>> Will wait db replication_db on localhost:5432, will try 50 times with delay 5 seconds
psql: could not connect to server: Connection refused
	Is the server running on host "pgstandby" (127.0.0.1) and accepting
	TCP/IP connections on port 5432?
psql: could not connect to server: Connection refused
	Is the server running on host "localhost" (::1) and accepting
	TCP/IP connections on port 5432?
could not connect to server: Connection refused
	Is the server running on host "localhost" (127.0.0.1) and accepting
	TCP/IP connections on port 5432?
...

Any help would be grealty appreciated.

EDIT:
The docker run for pg2 was wrong. I fix it.

EDIT2:
The first server is running and listening:
$ netstat -an | grep 5432
tcp6 0 0 :::5432 :::* LISTEN

[jmcollin78]

When looking to entreypoint.sh I see that Postgresql in not started in case of not master server:

if [[ "$INITIAL_NODE_TYPE" == "master" ]]; then
    cp -f /usr/local/bin/cluster/primary.entrypoint.sh /docker-entrypoint-initdb.d/
    /docker-entrypoint.sh "$@" &
else
    /usr/local/bin/cluster/standby.entrypoint.sh "$@" &
fi

Postgres is started by the /docker-entrypoint.sh shell and it is not called in case of slave server.
Do I miss something ?

[jmcollin78]

OK, I found what is going wrong. My Pg2 container must be linked to the Pg1 as pgmaster. The correct docker run for Pg2 is:

docker run --rm --name Pg2 --link Pg1:pgmaster --env INITIAL_NODE_TYPE=standby --env NODE_ID=2 --env NODE_NAME=node2 --env CLUSTER_NODE_NETWORK_NAME=Pg2 --env REPLICATION_UPSTREAM_NODE_ID=1 --env FORCE_CLEAN=1 --env REPLICATION_PRIMARY_HO
ST=pgmaster --env REPLICATION_PRIMARY_PORT=5432 --env CLUSTER_NAME=pg_cluster --env REPLICATION_DB=replication_db --env REPLICATION_USER=replication_user --env REPLICATION_PASSWORD=replication_pass -p 5433:5432 -v /data/pg2:/var/lib/postg
resql/data paunin/postgresql-cluster-pgsql