You must produce some code that reproduces this issue, if it only happens with Upstash then great, but we need to know what you are running so that we can have any possible chance to fix it.

Also, could you verify the exact version from which the issue can be reproduced? that may help in narrowing down the potential origin of the issue.

Can confirm v5.30.0 introduces the bug. The following code should suffice to reproduce.

import { Redis } from 'ioredis'
import { Queue, Worker } from 'bullmq'

const REDIS_URL = 'redis://default:[email protected]:6379'

export const connection = new Redis(REDIS_URL, { maxRetriesPerRequest: null })

async function main() {
	const myQueue = new Queue('myQueue', { connection })

	const worker = new Worker('myQueue', async (job) => {
		console.log('Processing job:', job.id)
		console.log('Job data:', job.data)
		return job.data
	}, { connection })

	await myQueue.add('testJob', { message: 'Hello BullMQ!' })
}

main().catch(console.error)

v5.30.0 only has a small change for job schedulers I cannot imagine that it would introduce an issue like this.

Sorry, it's v5.33.0

➜  cat package.json | jq '.dependencies'
{
  "bullmq": "5.32.0",
  "ioredis": "^5.3.2"
}
➜  node index.js
Processing job: 1
Job data: { message: 'Hello BullMQ!' }

➜  pnpm i [email protected]
Packages: +1 -1
+-
dependencies:
- bullmq 5.32.0
+ bullmq 5.33.0 (5.34.1 is available)
➜  node index.js
Processing job: 2
Job data: { message: 'Hello BullMQ!' }
Error: Missing lock for job 2. moveToFinished

Could you possibly to a "redis-cli monitor" against the upstash instance when you run the test that reproduces the issue? The output of the monitor would be very useful to find the issue.

Here it is redis-monitor.csv:

Thanks, this was useful.

I have a fix that I hope will solve this for you. The reason I think this fails in Upstash is because they use a different LUA engine with a different version of msgpack.

Please try again with v5.34.2