Setting up cluster
Before you set up the nodes in a cluster you should have already configured
a cache server, as described in the high-level steps below. The cluster requires the
presence of a cache server in order to cache data that should be available to all nodes
in the cluster.
If your cache server isn't configured and running, you won't
be able to set up the cluster.
Note: Your license determines whether or not clustering is enabled and how many nodes are
supported. To check on the number of clustered servers your license allows, see the
license information after logging into the admin console.
Topology
The nodes in a cluster need to be installed on the same subnet, and preferably on the same switch. You cannot install nodes in a cluster across a WAN.
Important notes
Before you get started, let me show start and related to setting up a cluster.
Important: If you're upgrading and copying the home directory (such
as
/usr/local/jive/applications/<instance_name>/home
) from
the older installation, you must preserve the node.id file and the crypto directory
from the home directory before starting the server. The value stored in this file
must be unique among the cluster nodes; that is, each node in a cluster must have a
unique value in the node.id file. You must preserve the node.id file because it
plays a role in storing encrypted information in the cluster; if that file is lost,
you can lose access to the encrypted information.- If you are deploying a new cluster, it is permissible to copy the contents of the home directory from the first node (where you set up clustering) to subsequent nodes — with the exception of the node.id file. Do not copy the node.id file to subsequent nodes. If the node.id file does not exist, the application generates a new file on startup.
- Always wait for the first node in the cluster to be up and running with clustering enabled before you start other cluster nodes. Waiting for a minute or more between starting each node ensures the nodes are not in competition. As the senior member, the first node you start has a unique role in the cluster. For more information, see Clustering in Jive.
- The cache server must be cleared and restarted before upgraded application server nodes are started and try to talk to the cache.
- If you're upgrading a plugin, clear the cache for that plugin and shut down the cache server first.
- The clocks on all machines must be synchronized in order for caching to work correctly. For more information, see Managing in-memory cache servers. Also, if you're running in a virtualized environment, you must have VMware tools installed in order to counteract clock drift.
- If you're running in a virtualized environment, you must have VMware tools installed in order to counteract clock drift.
- If your deployment places a firewall between cluster nodes and cache servers, be sure to leave the following ports open between machines: 6666, 6667 and 6650. Caching won't work correctly unless these are open.
- Port 6650 should be blocked to external access (but not between the cluster nodes!) so that any access outside of the data center is prohibited. This is to prevent operations allowed by JMX not to be executed on the cache server.
High-level new installation steps
Important: If, as part of your new installation, you're setting up one node
as a template, then copying the home directory (such as
/usr/local/jive/applications/<instance_name>/home
) to other
nodes in the cluster, you must remove the node.id file and the crypto
directory from the home directory before starting the server. The application will
correctly populate them.To set up a cluster:
- Use the application package (such as the RPM on Linux) to set up a cache server on a separate machine. For more information, see Setting up cache server. Note the cache server address for use in setting up application servers.
- Before proceeding, make sure the cache server you set up is running. It must be running while you set up application server nodes.
- On each node in the cluster, install the application instance using the package
(RPM on Linux or package on Solaris), but don't run the Setup wizard yet.
For the installation instructions about installing the application with a package, see Installing Jive package and starting up.
- Start one node and navigate to its instance with a web browser. In the setup screen provided, enter the address of the cache server you installed, then complete the Setup wizard.
- After you've finished with the Setup wizard, restart the application server.
- Manually copy the encryption keys from the
/usr/local/jive/applications/app_name/home/crypto
directory to each of the other nodes. - Start the application server on each of the other nodes. Because it's connecting to the same database used by the server on the node you've already set up, the server on each subsequent node detects that clustering is enabled. Each also picks up the configuration you set on the first node.
- After setting up all of the application server nodes and running them once, restart all servers in the cluster to ensure that the address of each node in the cluster is known to all the other nodes. The entire cluster must be bounced after all the nodes are set up.