High Availability Licensing
High-availability licensing enables you to set up an environment in which there is a very high degree of certainty that concurrent licenses will be available, even if a network license server goes down.
A cluster is a set of active servers configured to communicate with each other through direct binding mode. In this mode, several workstations serve licenses and others are in reserve ready to take over from an unavailable server.
For more information about clusters, refer to the manual Using License Use Management Runtime for your platform: section "Planning Clusters for High-Availability Licensing" in Chapter 2 "Planning Network Licensing."
For more information about configuring clusters, refer to the manual Using License Use Management Runtime for your platform: section "Administering High-Availability Licensing (HAL)", "Scenario 11: Creating and Managing a Cluster" in Chapter 4 "Administering License Use Management Runtime".
Note that if you wanted to set up a mixed Windows and UNIX cluster, LUM 4.5.9 or higher is required.
However, you should be aware of the following situation which might occur when using HAL clusters.
|In this example, let's suppose you have one license on a 3-server HAL
cluster. A first CATIA client requests the license, and the license is
granted by the first server.
Then, the first server shuts down.
The next time that CATIA connects to the first license server to check that it is still up and running, CATIA displays several warning messages because it is not certain that the license can be acquired or not on another server in the cluster.
The license is then granted by the second server and CATIA goes on.
After the first server shuts down, another CATIA client on another machine may request the same license from the first server which has shut down. The same license is then granted to this client by the second server. The next time that CATIA connects to the first license server to check that it is still up and running, CATIA displays several warning messages on the first CATIA client then exits, because the license is already granted to the second CATIA.
This problem is due to the fact that the second server knows that the first server is down, but doesn't take into account the licenses granted by the first server. If this was the case, the second CATIA process wouldn't start, the same license wouldn't be granted to two CATIA processes, and the first CATIA process wouldn't exit.
In both cases, the messages which appear are insignificant, and should be ignored.
These problems are due to the way in which HAL was originally designed.
Servers that are members of a HAL cluster share license availability information, but not license usage information. Consequently, the other servers do not know that a certain license was in use on the server that went down, consider this license as available, and grant it to the first client that requests it.
HAL was not designed to share this type of license usage information among all the servers in a cluster.