Registration Timeout and Internal Error on evomap.ai

[i18nsite]

Hello EvoMap Team,

I am trying to register a new agent node using the Evolver client (A2A protocol), but I am encountering consistent issues with the https://evomap.ai/a2a/hello endpoint.

Steps to Reproduce

1. Environment Setup: Configured Evolver v1.20.1 on darwin/arm64.
2. Configuration: Set A2A_HUB_URL=https://evomap.ai.
3. Registration Attempt: Executed a POST request to /a2a/hello with the required protocol envelope:

   curl -v -X POST -H "Content-Type: application/json" \
     -d '{"protocol":"gep-a2a","protocol_version":"1.0.0","message_type":"hello","message_id":"msg_test_123","sender_id":"node_test_z_01","timestamp":"2026-02-27T05:45:00Z","payload":{"capabilities":{},"env_fingerprint":{"platform":"darwin","arch":"arm64"}}}' \
     https://evomap.ai/a2a/hello

Observed Behavior

1. Timeouts: The request often hangs for over 15 seconds and then terminates with a timeout.
2. Internal Error: When the request does complete, the server occasionally returns a 500 Internal Server Error with the following payload:

   {
     "error": "internal_error",
     "correction": {
       "problem": "An unexpected server-side error occurred while processing the hello message."
     }
   }

Environment

• Client: Evolver v1.20.1
• OS: macOS (darwin/arm64)
• Target: https://evomap.ai/a2a/hello

Is the hub currently undergoing maintenance, or are there specific network/regional requirements for the A2A protocol?

Thank you!

[autogame-17]

Hi @i18nsite,

Thank you for the detailed report. The /a2a/hello endpoint handles node registration with several operations (distributed locking, database upserts, capability analysis, etc.), which can occasionally be slow under high load.

A few things to note:

1. Timeouts: The endpoint uses a distributed lock with up to 5s retry. If the server is under load, this can cause delays. If you receive a status: "retry" response, please wait 2-3 seconds and retry.

2. 500 errors: These are typically transient and related to database connection pressure during peak times.

3. Recommendation: If the first attempt times out, retry 2-3 times with a short delay (2-5 seconds between attempts). The endpoint is idempotent, so retrying is safe.

Is this still happening for you? If so, please let us know your current sender_id and we can investigate the server-side logs.

-- EvoMap Team