Tip:
Highlight text to annotate it
X
Configuring a cluster: first node, additional nodes, disaster recovery node.
Using a disaster recovery node.
The entire cluster appears as a single server to the user.
But initially, each node is configured individually.
Every cluster starts with a first node.
Let's configure the first node.
For every node, you specify during the installation
that this server will be part of a cluster.
(Except for the disaster recovery node.)
The client download page does not display
until you complete the node configuration.
You need to reserve an IP address on your network
to be used as virtual address of the cluster.
Correct time settings are crucial for cluster nodes.
A reliable NTP server is recommended.
You can improve cluster security
by specifying a range of IP addresses
allowed to administrate the cluster.
Configuring the node takes time
and starts the application server.
Configuring other nodes
during the startup is not recommended.
Wait for the first node to complete the startup.
We now have a cluster, but with only one node.
Clients will connect to the virtual IP address of the cluster,
not the IP of the node.
Let's add more nodes to the cluster.
Log in to Webmin of the first node.
First, you need to tell the cluster
to allow the new IP of the next node.
The new node should be running at this point.
You should verify the fingerprint of the new node.
Note that here you jump to the new node.
Now, tell the new node which cluster to join and specify the cluster's IP.
Again, you should verify the fingerprint.
Note that this is the fingerprint of the cluster.
The IP address and fingerprint of the cluster
are remembered by the node.
This prevents "middleman attacks" on the cluster.
The new node will be synchronised with cluster data
and it's status will change to "Running".
With the second node running your data becomes fully redundant.
Let's add a disaster recovery node.
A DR node is initially installed as standalone.
When the DR node is started, return to the Webmin cluster setup.
In the Webmin cluster setup, tell the cluster
to allow the new IP of the DR node.
Again, you should verify the fingerprint of the new node.
Note that here you jump to the DR node.
Now, you tell the DR node to join the cluster.
The new node just needs to know the virtual IP of the cluster.
After connecting to the cluster,
the DR node begins to receive the updates.
Let's review our cluster setup in Webmin.
Here, you can see the list of nodes in the cluster.
Note the status of each node.
We are done with cluster configuration.
You can add more nodes to a cluster,
further improving performance and availability.
Under normal conditions,
clients communicate with the cluster.
If the entire cluster is disabled,
you can promote the recovery node to a standalone server.
Clients can work with it,
although with reduced performance.
When the cluster is available again,
you can restore data from the DR node.
This is a regular backup and restore procedure.