Scenario: Knowing When You Need More Proxiesedit

The following scenario shows you how you can identify the CPU and memory usage of a runner that is holding the proxy role and is routing user requests to Elasticsearch and Kibana. If these usage indicators are consistently high, they might indicate that you need to add another proxy to your Elastic Cloud Enterprise installation or when it is time to separate the proxies from other parts of your installation, such as our medium-sized playbook example.

To find the CPU and memory usage for proxies:

  1. Access the logging and metrics information in Kibana.
  2. Open the Dashboard app and select Open.
  3. Select the Processes dashboard and search for proxy.

    The results include process information for java, internal Docker proxies, and runsv. The CPU and memory usage for the runners that hold the proxy role is indicated by the java system process. Alternatively, you can search for system.process.name:java AND proxy to show only the right information.

    CPU and memory usage of proxies

    Usage is shown per runner. As you can see, the CPU and memory usage of the proxies is quite low and within single digits. One runner shows less information than the other, because it was only recently assigned the proxy role to improve fault tolerance. Note that for fault-tolerant installations, the remaining proxies must be able to handle the volume of user requests if a proxy fails. In this case, the proxies are over-provisioned and the failure of one proxy will not affect the ability of Elastic Cloud Enterprise to route user requests to Elastisearch and Kibana.