BIG-IP heap tuning with AI Chat

Overview

F5 Insight AI Chat can check and apply restjavad and tomcat heap settings on BIG-IP devices. This is useful when you investigate restjavad out-of-memory (OOM) errors or slow iControl REST responses.

Note

Unlike other AI Chat tools that are read-only, this feature can modify BIG-IP device settings. Use it with caution.

For background on restjavad memory provisioning, see K000137363: Overview of restjavad memory provisioning on BIG-IP.

Before you start

  • AI Chat must be enabled with a configured LLM provider. See LLM insights.
  • You must be logged in as an Admin or Operator. Viewer roles can’t modify BIG-IP device settings.
  • The target BIG-IP device must be onboarded and reachable from F5 Insight.

How it works

AI Chat provides two operations for heap tuning:

Check heap status

Ask AI Chat to check the current heap allocation on a device or all devices in a datacenter. AI Chat compares the current values against recommended settings and flags any devices that are undersized.

Example query

Check the heap status on device bigip-01

Apply heap tuning

Ask AI Chat to apply the recommended heap settings on a specific device. This operation:

  • Sets provision.extramb to the recommended value (2048 MB).
  • Sets provision.tomcat.extramb to the recommended value (256 MB).
  • Restarts the restjavad service.

Warning

Applying heap tuning causes a 30-60 second iControl REST outage on the target device while restjavad restarts. AI Chat asks for explicit confirmation before it proceeds.

Example query: “Apply heap tuning on device bigip-01”

Verify the result

After AI Chat applies the heap tuning, it reports the updated values. You can also verify by asking:

Example query: “Check the heap status on device bigip-01”

The response shows the updated provision.extramb and provision.tomcat.extramb values. Both should match the recommended settings (2048 MB and 256 MB).

Note

The restjavad service takes 30–60 seconds to restart. Wait for the restart to complete before you verify the heap status or send iControl REST requests to the device.

Adjusted values for memory-constrained devices

If the target BIG-IP device doesn’t have enough memory for the recommended heap settings, AI Chat automatically reduces the values to fit within the device’s available memory. In this case, the response indicates the adjusted values instead of the defaults.

Example response for a memory-constrained device:

The originally recommended 2048 MB / 256 MB values exceeded the device's
hardware capacity.

Applied values were reduced to 512 MB / 64 MB to stay within device memory
limits.

This optimization should help prevent restjavad OOM errors and improve
iControl REST performance.

The device had memory limits of 7984 MB available vs 8290 MB+ required
for larger allocations.

Important

Even with reduced values, the adjusted heap settings improve restjavad stability compared to the default allocation. If OOM errors persist after tuning, consider upgrading the device’s memory or reducing the number of provisioned modules.