can1357/oh-my-pi · info · HTTPException

replay disabled

Error message

replay disabled

What it means

The /replay endpoint re-dispatches a stored webhook delivery, but only when ROBOMP_REPLAY_TOKEN is configured. If cfg.replay_token is None the endpoint raises 404 'replay disabled' — replay is opt-in and off by default.

Source

Thrown at python/robomp/src/server.py:508

            pool: WorkerPool = bag["pool"]
            pool.wake()
            log.info(
                "queued", extra={"event": x_github_event, "delivery": x_github_delivery, "key": decision.issue_key}
            )
        else:
            log.info("duplicate", extra={"event": x_github_event, "delivery": x_github_delivery})
        return JSONResponse({"delivery": x_github_delivery, "state": "queued"}, status_code=202)

    @app.post("/replay")
    async def replay(
        request: Request,
        x_robomp_token: str | None = Header(None, alias="X-Robomp-Replay-Token"),
        delivery_id: str = "",
    ) -> JSONResponse:
        bag = request.app.state.bag
        cfg: Settings = bag["settings"]
        if cfg.replay_token is None:
            raise HTTPException(404, "replay disabled")
        if x_robomp_token != cfg.replay_token.get_secret_value():
            raise HTTPException(401, "invalid replay token")
        db: Database = bag["db"]
        row = db.get_event(delivery_id)
        if row is None:
            raise HTTPException(404, "unknown delivery")
        if not db.requeue_event(delivery_id, from_states=INACTIVE_EVENT_STATES):
            raise HTTPException(409, f"delivery {delivery_id} is {row.state}; only inactive events can be replayed")
        bag["pool"].wake()
        return JSONResponse({"delivery": delivery_id, "state": "queued"})

    def _require_trigger_token(cfg: Settings, token: str | None) -> None:
        if cfg.replay_token is None:
            raise HTTPException(404, "trigger disabled (set ROBOMP_REPLAY_TOKEN to enable)")
        if token != cfg.replay_token.get_secret_value():
            raise HTTPException(401, "invalid replay token")

    @app.get("/api/github/issues")

View on GitHub (pinned to 9690622007)

Solutions

1. Set ROBOMP_REPLAY_TOKEN in the orchestrator environment and restart to enable replay.
2. Call the endpoint with header X-Robomp-Replay-Token matching the configured token.
3. If replay should stay disabled, use an alternative recovery path (e.g. redeliver the webhook from GitHub's delivery pane).
4. Check the CLI output/HTTP status distinguishes 404 'replay disabled' from 401 'invalid replay token' to confirm which problem you have.

Example fix

// before (.env)
# replay not configured
// after (.env)
ROBOMP_REPLAY_TOKEN=<long-random-secret>

import os
if not os.environ.get('ROBOMP_REPLAY_TOKEN'):
    raise SystemExit('Replay is disabled: set ROBOMP_REPLAY_TOKEN first')

try:
    resp = httpx.post(f'{base}/replay/{delivery}', headers={'X-Robomp-Replay-Token': token})
except httpx.HTTPStatusError as e:
    if e.response.status_code == 404:
        ...  # replay disabled (404) vs invalid token (401) vs unknown delivery (404)

When it happens

Trigger: POST /replay (or `robomp replay <delivery_id>` against a server) when the Settings has no replay token configured.

Common situations: Operator running `robomp replay` in a deployment where the env var was never set; fresh install from .env.example without the optional replay token; intentionally hardened production without replay.

Related errors

• Devin AssignModel error ${response.status} ${response.status
• GitLab Duo Workflow create failed with HTTP ${response.statu
• ${response.status} ${response.statusText}: ${text}
• the discovery endpoint returned incomplete upload-node metad
• Destination option ${optionName} must use http or https

AI-assisted analysis of can1357/oh-my-pi@9690622007 (2026-08-31). Data as JSON: /api/errors/d0ec9f51d06ff0e7. Report an issue: GitHub.