Skip to content

Agent monitoring

Every load generator running during your test is automatically monitored.

You can get the metrics from any line chart or by using one of the shorcuts below:

Agent monitoring

Warning

Activating IP Spoofing from one of your on premise load generators will deactivate the agent monitoring because of the changes made to networking.

Operating system monitoring

The list of available counters on the Linux host and their alert levels is as follows:

OS monitoring

  • CPU:

  • % CPU Usage: % of CPU used: (100 - CPU Idle),

  • Memory:

  • % Used Memory: % Effectively Used memory: (used - (buffered + cached) / total * 100).

  • Network:

  • Received MB/sec: Number of MegaBytes per second received,
  • Sent MB/sec: Number of MegaBytes per second sent,
  • Established connections: Number of established connections.

  • TCP:

  • % Segments Retransmitted: Percent of TCP segments retransmitted.

For example here's a graph showing an alert with the Segments retransmitted:

Retransmit

This indicates a network contention on the remote server being tested.

JVM monitoring

The list of available counters on the Java Virtual Machine running JMeter and their alert levels is as follows:

JVM monitoring

  • JVM:
  • Memory:

    • used: memory used by the JVM in bytes,
    • max: max JVM memory in bytes,
    • % used: Percent used memory (used / max).
  • Garbage Collector:

    • G1 Young Generation:
    • CollectionCount: Number of garbage collections,
    • CollectionTime: Time spend in garbage collection.
    • G1 Old Generation:
    • CollectionCount: Number of garbage collections,
    • CollectionTime: Time spend in garbage collection.
  • Threading:

    • ThreadCount: Number of threads currently running in the JVM,
    • DaemonThreadCount: Number of daemon threads currently running in the JVM,
    • PeakThreadCount: Maximum number of threads running since JVM startup.

The most common issue is with the JVM Heap space. Typically when you run out you would see something like this:

Heap

That means the Heap memory allocated for your test was not enough. We already fine tune the JVM and use large cloud instances to avoid most issues but when running a low amount of users or resource intensive scripts, you might need to do one of the following:

Be careful not to confuse Heap space issues with CPU issues since an insufficient Heap space will also result in a high CPU usage:

heap-cpu

Typically here, whenever the G1 Old generation garbage collection starts, the load generator is quickly out of CPU.