Related Issue

Closes #1964

Summary

This PR fixes an issue where healthy but slow gbrain engines could be incorrectly classified as broken-config when the probe operation exceeded the default timeout threshold.

In environments where the gbrain CLI takes longer than 5 seconds to respond (for example, when using remote databases or connection poolers), the probe would time out and fall back to the generic broken-config classification. As a result, /sync-gbrain skipped the code and memory stages even though the engine itself was healthy and operational.

Changes Made

1. Added a dedicated slow-engine status

Introduced a new LocalEngineStatus value:

"slow-engine"

This allows timeout-related failures to be distinguished from actual configuration problems.

2. Improved timeout handling

The classifier now explicitly detects timeout conditions (such as ETIMEDOUT) and returns slow-engine instead of incorrectly falling back to broken-config.

3. Made probe timeout configurable

Replaced the hardcoded timeout value with an environment-variable-based configuration:

GSTACK_GBRAIN_PROBE_TIMEOUT_MS

This allows users with slower database connections or remote infrastructure to adjust the probe timeout without modifying source code.

4. Increased the default timeout

Updated the default probe timeout from:

5000ms

to:

20000ms

This better accommodates slower but healthy environments while still preventing indefinite waits.

5. Added a user-friendly sync message

When a slow engine is detected, users now receive a clear explanation indicating that the engine may still be healthy and suggesting increasing the timeout or improving database connectivity, rather than being told their configuration is malformed.

Why This Change?

The original behavior treated timeout failures as configuration failures, which could mislead users into troubleshooting the wrong component. This PR separates slow-response scenarios from actual configuration issues, making diagnostics more accurate and improving the overall user experience.

Testing

• Verified that timeout-related failures are classified as slow-engine.
• Verified that existing broken-config and broken-db classifications remain unchanged.
• Verified that the timeout can be configured using GSTACK_GBRAIN_PROBE_TIMEOUT_MS.
• Verified that sync skip messages now provide more accurate guidance for slow environments.

Checklist

• Code follows project style
• No unrelated changes included
• Existing behavior preserved for valid configurations
• Added clearer handling for slow-engine scenarios