Updated Date: 10/27/2026
Health Engine Overview
The health engine is a diagnostic framework built into F5OS. It runs a set of diagnostic scripts, called heuristics, directly on the platform to identify known issues and provide actionable health recommendations. This reduces diagnostic time by eliminating the need to generate, transfer, and upload a qkview to iHealth.
The health engine performs structured analysis of system data on the device itself. Each heuristic script checks for a specific condition, evaluates the result, and can recommend an action for you to take.
Traditional diagnostics with qkview can take a long time. The capture process on a chassis can take hours. Transferring the file and uploading it to iHealth adds more delays. A technical expert often needs to review the data before you get answers.
The health engine addresses these limitations:
- Faster time to diagnosis: Analysis happens on the platform, so you skip the capture-transfer-upload cycle.
- Actionable recommendations: Heuristic scripts assess conditions and suggest specific steps, such as restarting a service or capturing a qkview.
- Background monitoring: The engine can run heuristics continuously in the background and report issues before they affect your operations.
The health engine does not replace qkview. You still use qkview when deeper analysis is needed. The engine provides a first line of diagnostics so you can resolve common issues faster.
The health engine supports three primary use cases:
As an F5 customer, you can run a set of heuristics on the F5OS platform to perform first-line diagnostic procedures and health checks. These are the same types of checks that a technical expert would normally perform by analyzing qkview data. Running them directly on the platform gives you faster results.
As an F5OS service, the health engine can run heuristics in the background. This generates actionable information continuously, so you can identify issues before they affect your operations.
As a diagnostic agent, the engine can trigger heuristics based on system events. Examples of trigger events include:
- Firmware update complete
- Blade powered on
- Other pre-defined events of interest
Note: The system-observer component, which triggers heuristics based on system events, is not included in F5OS 1.8.3-C or 2.1.0. This capability is planned for a future release.
The engine follows a three-step process:
- Load: The engine discovers and loads heuristic scripts based on file type, location, or filtering criteria such as tags or platform role.
- Run: The engine invokes each script through a defined entry point. Scripts inspect live system data by accessing interfaces such as the F5OS confd API, the file system, platform services, and the tmstat database.
- Report: Each script produces results that include log messages, conditions found, recommended actions, and files. The engine collects individual results and writes them to result files.
The health engine is one of several diagnostic approaches available in F5OS:
| Approach | Description |
|---|---|
| Interactive | You manually inspect the system through the confd CLI, web UI, or tmsh. |
| Qkview deep analysis | A large data set is captured and uploaded to iHealth for expert analysis. |
| Structured analysis (health engine) | Diagnostic scripts run on the platform and return results and recommendations. |
Each heuristic script has two main parts:
- Metadata: Describes the script, including its unique ID, name, description, supported platform roles, filtering tags, and trigger events. The engine reads this metadata to decide which scripts to load and run.
- Entry point: Contains the diagnostic logic. This is where the script inspects system data, evaluates conditions, and generates results.
Scripts produce four types of output:
- Log messages: Internal diagnostic information used by F5 for troubleshooting. These are not directly visible to you.
- Results: Conditions the script has detected, such as a mismatch between actual and expected hardware speeds. Each result has a severity level based on standard syslog levels.
- Actions: Recommended steps for you to take, such as restarting a service or capturing a qkview. Some actions run automatically. Actions that could affect system performance are recommendations only and require you to act on them.
- Files: Files written to the logging directory by the script during execution. These can be used for additional diagnostic data capture.
The engine provides health recommendations through an API. Recommendations tell you what action to take, which component is affected, and how severe the issue is.
Heuristic scripts can recommend the following actions:
| Action | Automated or recommended |
|---|---|
| Restart a node (blade, controller, or appliance) | Recommended |
| Shut down a node (blade, controller, or appliance) | Recommended |
| Capture a qkview | Recommended |
| Restart a service | Recommended |
| Generate or clear an alarm | Automated |
| Generate an event | Automated |
| Adjust logging level | Automated |
| Update diag-agent component health | Automated |