Start and stop servers in a cluster
The procedures used to start and stop the primary node in a cluster are identical to the procedures used to start and stop the PPM Server in a single-server configuration. (For detailed information, see Start and stop the PPM Server on a single-server system.)
Caution: If your PPM instance includes multiple nodes in a cluster configuration, you must start these nodes one at a time. Make sure that you wait until each node is fully started before you start the next node.
To start a secondary node, use the
-name server-name argument in the
kStart.sh script, as follows.
sh ./kStart.sh -name <PPM Server>
To stop a secondary node, run the
kStop.sh script, as follows:
sh ./kStop.sh -name <PPM Server> -now -user <User_Name>
On Windows, there is one service (called "
PPM <PPM Server>") per node. If you prefer to use the Windows shell command line to start nodes instead of using Windows Services, you can use the
If you do not have a script to stop all nodes in a cluster, you can write a script for this purpose. The following example script for the UNIX environment stops all three nodes in a cluster configuration (all nodes are on the same machine).
./kStop.sh -name serv1 -now -user <User_Name>
./kStop.sh -name serv2 -now -user <User_Name>
./kStop.sh -name serv3 -now -user <User_Name>
A PPM Server cluster continues to operate as long as at least one node in the cluster is running. If a node stops, the PPM web server module detects that the node is unavailable and stops sending it HTTP requests. When the node becomes available again, the PPM web server module detects the node and sends the requests again.
Note: If you make a change to the
server.conf file that affects more than one node in a cluster, you must:
Stop and restart (one at a time) all the nodes in the cluster.
kUpdateHtml.shscript on all machines.