Note: This was originally posted by an inactive account. Content was preserved by moving under an admin account.
Originally posted by: Tim MeagherHi,
Sorry you're experiencing this problem.
While you may have enough available memory on the system, there is not enough contiguous memory available for the Java Virtual Machine (JVM) at the time of launching.
Since the LAE server is a Java application, it relies on the JVM being able to get enough memory to be able to launch.
The problem is that in order to specify a sensible maximum limit to the amount of memory the server can use, the maxHeapSize setting must be used.
Unfortunately, this memory has to be available (though it will not be used unless the server requires it) to the JVM at the time of launching, and needs to be contiguous memory space available in an area of the memory space accessible by the JVM.
Now... the server is configured to have a maxHeapSize of 1024M (or 1GB).
However, in general - unless you are running very large graphs using the external controller, running large command line BRX's, or using an awful lot of drones it won't need this much memory.
So in order to work around this problem, there are two options you can look at.
The first and probably easiest is to simply shut down unused applications, or things that you don't need
right now then restart the server.
Once the server is up and running you restart those applications again without problem.
The only problem with this approach is you may need to do this whenever you go to start the LAE server.
The alternative is to go into your LAE installation directory, and edit the conf/site.prop.
At the end of the site.prop file, you can add the properties:
ls.brain.server.javaMaxHeapSize=<SomeValue>
Where <SomeValue> can be set to a value that is reasonable for your machine.
For instance, to set the server to have a maximum of 512MB memory available you can set this to:
ls.brain.server.javaMaxHeapSize=512M
The Java based nodes within the LAE also need to launch a JVM so may run into the same problems.
If you experience such problems running Java based nodes (such as the Java node itself, or the Excel, Output Excel, Pivot nodes, or JDBC nodes - among others) then you may also need to configure the maximum heap size available to the Java based nodes as well.
You can do this using a similar mechanism in the site.prop file, by adding a line like:
ls.brain.node.java.javaMaxHeapSize=<SomeValue>
e.g.
ls.brain.node.java.javaMaxHeapSize=512M
Bear in mind that while the server normally doesn't require that much memory as it is simply coordinating the nodes and the requests from BRE, the nodes themselves which do the processing will often require more memory than the server.
Hopefully this will solve the problem for you, however for more information and assistance, or if these suggestions don't help you address the problem, you should contact your Lavastorm Support representative.
Regards,
Tim.