Tasks are currently stuck in queue without no machine attached

[ember crescent]

Hi @feral sparrow. I'm kapa.ai, your friendly support bot. I will try to answer your question by searching through the knowledge base. :wave:

---

Thanks for flagging this—let’s run through the likely causes and quick checks based on our docs:

• Environment concurrency/queue limits: Only actively executing runs consume concurrency slots; runs in WAITING or queued state don’t. If your environment or a shared queue is at its concurrency limit, new runs will sit queued until a slot frees up. Verify any custom queues’ concurrencyLimit and whether many runs are waiting at checkpoints. [Concurrency & Queues; Waits and concurrency]

• Versioning/deployment: In Staging/Production, scheduled tasks only trigger from the latest deployment; older deployments won’t run. Make sure you’ve deployed and that runs are on the current version. [Scheduled tasks; Versioning]

• Machines: By default, tasks run on small-1x unless overridden. You can set a task-level machine or a default in trigger.config.ts. If you replay with a different machine but the run stays queued, it still must acquire concurrency on its queue/environment. Consider raising the machine only if you’re hitting OOM—not for queueing. [Machines; Default machine; Limits: Machines]

• Global limits: Check your plan’s concurrency and rate limits. If you’re batch triggering a lot at once, ensure you’re not rate-limited. [Limits]

Troubleshooting steps:

1. Check queues and concurrency:

• Confirm if your tasks specify a shared queue and what its concurrencyLimit is.
• If needed, temporarily increase concurrencyLimit on that queue. [Concurrency & Queues]

2. Confirm deployment/version:

• Ensure you’ve deployed the latest version and that runs are locking to it. [Versioning]

3. Verify environment:

• For Dev, make sure the dev CLI is running; for Staging/Prod, confirm tasks exist in the current deployment. [Scheduled tasks]

4. Use MCP tools to inspect/trigger:

• List tasks and trigger a small payload in the affected environment to test: [MCP tools]

{
  "tool": "get_tasks",
  "arguments": { "projectRef": "proj_abc123", "environment": "prod" }
}

{
  "tool": "trigger_task",
  "arguments": {
    "projectRef": "proj_abc123",
    "environment": "prod",
    "taskId": "your-task-id",
    "payload": "{}",
    "options": { "machine": "small-2x" }
  }
}

If these checks don’t surface the issue, the knowledge sources don’t include a documented cause specific to “no machine attached” on v4.0.0-v4-beta.27. Please share:

• Task IDs/run IDs,
• Environment (dev/staging/prod),
• Queue config (names and concurrencyLimit),
• Whether many runs are in WAITING vs RUNNING,
  and we’ll investigate. You can also reach us directly for urgent help. [Limits]

---

Note: you can ask me a follow up question by @ mentioning me again :speech_balloon:

@feral sparrow Kapa.ai is still learning and improving, please let me know how I did by reacting below.