Connecting to your PGD cluster
==============================

EDB Postgres Distributed (PGD) uses the same connection methods as
Postgres. The difference is that most of your connections to the cluster
go through the connection manager that’s built into every data node in
the cluster.

Although you can connect directly to the data nodes, we don’t recommend
it for anything other than maintenance when you want to work on a
particular node’s database instance.

In general, you must connect to the cluster through the connection
manager. PGD is designed to be simple to deploy and manage, and that
means the cluster has a write leader node that handles all the writes to
the cluster. The connection manager is then responsible for directing
your read-write connections to the write leader. All your client or
application needs to do is to use the connection manager’s port and the
connection manager will handle the rest.

The connection manager is responsible for directing your writes to the
write leader and ensuring that your reads are directed to the correct
node in the cluster. If you connect directly to a data node, you may not
be able to take advantage of these features. For applications that only
need to read data, the connection manager can direct your reads to a
node that isn’t the write leader. This can help to balance the load on
the cluster and improve performance.

Connecting through the connection manager
-----------------------------------------

Postgres is very flexible for configuring ports and connections, so for
simplicity, this example uses the default port settings for Postgres and
the connection manager. The default port for Postgres is 5432, and the
default port for the connection manager is 6432.

You can use that port in your connection strings to connect to the
cluster. So, for example, if you’re using the psql command line tool,
you can connect to the cluster like this:

.. code:: bash

   psql -h host-1 -p 6432 -U pgdadmin -d pgddb

Where ``host-1`` is the hostname of the node you’re connecting to. The
connection manager will then direct your connection to the write leader
node in the cluster.

Connecting directly to a data node
----------------------------------

You can connect directly to a data node in the cluster, but we don’t
recommend it. However, if you need to connect directly to a data node,
you can use the following command:

.. code:: bash

   psql -h host-1 -p 5432 -U pgdadmin -d pgddb