Prysm vc `500: UNHANDLED_ERROR: UnableToReadSlot`

[barnabasbusa]

Description

So as of 9th of dec 2025 the validator matrix was complete.
https://github.com/ethpandaops/assertoor-test/actions/runs/20048774613#summary-57500644627
Then on the 10th of dec 2025, the validator matrix broke for lighthouse CL <> Prysm VC. Prysm vc works with every other CLs, so I'm guessing something lighthouse specific has happened between the 9th and 10th of dec.

Version

All tests are running latest trunk branches of prysm and lighthouse too.

Present Behaviour

Prysm vc fails to start up with [2026-01-05 10:58:45.51] ERROR client: Could not create validator runner error=failed to update proposer settings: HTTP request unsuccessful (500: UNHANDLED_ERROR: UnableToReadSlot)

This looks like a Lighthouse bug - the prepare_beacon_proposer endpoint should return a 503 "Service Unavailable" (or retry-able error) when the slot clock isn't ready, not a 500 "Internal Server Error".

Expected Behaviour

We should have no issues starting up prysm vc with lighthouse cl.

Steps to resolve

kt config:

participants:
  - cl_type: lighthouse
    cl_image: ethpandaops/lighthouse:unstable
    vc_type: prysm
    vc_image: ethpandaops/prysm-validator:develop

[jimmygchen]

I'm able to reproduce this with the config provided, I'll look into this.