Manage an active/active cluster
The following procedures describe how to perform management tasks for an active/active cluster setup.
The following topic provide the how-to instructions for managing an active/active cluster:
Related topics:
Monitor an active/active cluster
Cluster status is displayed in the Administration Tool.
- Select Operations > Cluster Management.
- The Cluster Management page is displayed.
For each Server in the cluster, the page lists its status (online or offline), its type (primary or secondary), and its host name or IP address. Online status means the server is running and communicating with the cluster. Offline status means the server has been stopped, has failed, or cannot communicate with the cluster.
The Bounce, Bounce All, and Synchronize All buttons do not appear on secondary servers. To bounce a secondary server locally, see Reload server configuration.
On a secondary server, the time of the last manual synchronization is reported. The timestamp is also reported in the cluster_last_sync_timestamp
file located in the <FILEDRIVEHOME>/var/tmp/cluster_last_sync_timestamp
directory.
Cluster status is also stored in the cluster_state
file. It contains information about which cluster node is the primary server, which are secondary servers, which servers are online, and which are offline. For example:
<ClusterGroup name="STCluster"> <Member hostname="test01.your.cluster" state="online" status="primary"/> <Member hostname="test02.your.cluster" state="online" status="secondary"/> <Member hostname="test03.your.cluster" state="offline" status="secondary"/> </ClusterGroup> |
By default, the cluster_state
file is located in <FILEDRIVEHOME>/var/tmp/cluster_state
. You can change the location and file name by editing the following server configuration parameters:
Cluster.File.clusterStateFile.relative
– the base location (fdhome
for <FILEDRIVEHOME>)
Cluster.File.clusterStateFile.path
– the path and file name relative to Cluster.File.clusterStateFile.relative
Add a server to an active/active cluster
- Stop the TM server on all servers in the cluster.
- Add the information about the new server into the
<FILEDRIVEHOME>/lib/admin/config/servers
file on the primary server.
- Copy the
servers
file to the new computer.
- Follow steps 11 through 15 from Set up an active/active cluster.
|
Note
|
The new server must use the same secret file as the rest of the nodes in the cluster. |
Restore a server to an active/active cluster
- Start the Administration Tool and TM server on the restored server.
- Perform a manual synchronization from the Administration Tool of the primary server. For instructions, see Synchronize the cluster from the primary server.
|
Note
|
Synchronize the restored server as soon as possible. If the restored server becomes primary before it is synchronized, it might process messages based on outdated data it received before it failed. |
Remove a server from an active/active cluster
When you remove a server from an active/active cluster, you can configure it as a stand-alone server.
- On each of the nodes in t he cluster, stop the TM server.
- On the node you are removing from the cluster:
- Open the
<FILEDRIVEHOME>/lib/admin/config/servers
file and remove the lines of information about the servers in the cluster. - On the Server Configuration page, set the
Cluster.mode
parameter to disabled
.
- On each of the remaining nodes in the cluster, open the
<FILEDRIVEHOME>/lib/admin/config/servers
file and remove the line of information about the server you are removing from the cluster.
- On each of the remaining nodes in the cluster, start the TM server.
- On the primary server, perform a manual synchronization from the Administration Tool. For instructions, see Synchronize the cluster from the primary server.