Admins can change the upper memory limit for any given service in the cluster using this function, which can be triggered from the Summaries Panel or the Server Instances page.
Changing Memory
The dialog simply requires a new value for the upper memory limit. This limit instructs the JVMs how much memory they can take for the service. In many situations, the maximum is not reached until the service is placed under pressure. Also, with Java, the memory is seldom reduced once it is taken. Any memory not actively required, is kept in the service's "heap" instead.
Clicking the Recommended Defaults button will generate a memory value from the system that best fits the deployment - taking the size of memory on the host server and the number of other services installed on the host.
Once the memory has been changed, the service will be restarted. During this period, the service is inaccessible.
Warnings
- RED: If the service being adjusted is the router, web server or runtime server and there is only a SINGLE instance of that service in the cluster a red warning will indicate that the entire cluster will be INACCESSIBLE while the service recycles.
- YELLOW: If the service being adjusted is the task server, AI server or file server a yellow warning will indicate that the service will be INACCESSIBLE while the service recycles. But the cluster will be operational otherwise. Similarly, a yellow warning will be shown if the service is the router, web server or runtime server and there are TWO or more instances of that service in the cluster, the cluster will remain operational.
Recommendations
Given that services (and possibly the entire cluster) are inaccessible during the adjustment, memory adjustments should be done in off-peak maintenance periods only.