sandboxed worker no timeout & how to fail job inside sandboxedworker

[arthurvanl]

I do not really care about how long a job is running. They can run as long as they need to process data. Is this a possibility?

I see this in the docs:

but I am not sure if this what I am searching for. I dont know what stalled jobs are in the full picture. Like I understand about the retries, but I dont use that. Every job has 1 attempt for me.

And how can I fail a job inside a sandboxedworker? The only function that the processor has is progress

How to use __flowParentId inside a sandboxed worker?

[egeominotti]

All three points addressed in v2.6.4 (now on npm):

• timeout: 0 now disables the timeout entirely for long-running jobs
• fail(error) is now available in the sandboxed processor to explicitly fail a job
• log(message) sends logs back to the parent process
• parentId is now exposed in the processor job object (from __flowParentId)

[arthurvanl]

Awesome update thank you a lot!

[arthurvanl]

1 More thing actually. The parentId I don't use. But I wanted to a get the parent result using the parentId

[egeominotti]

Hey! The reason I won't implement this is that in a flow, children are processed BEFORE the parent — so when your child job runs, the parent hasn't completed yet and has no result to fetch.

The intended pattern is the opposite: the parent uses getChildrenValues() to collect results from its children after they finish. If a child job needs data, it should be passed in job.data when creating the flow with FlowProducer:

const flow = await flowProducer.add({
  name: "parent",
  queueName: "main",
  data: { sharedConfig: "..." },
  children: [
    {
      name: "child",
      queueName: "tasks",
      data: { everythingYouNeed: "..." } // pass it here
    }
  ]
});

This way the child has all the context it needs without runtime lookups.

[arthurvanl]

My jobs are mostly sequential. So I keep using the FlowProducer#addChain. So that's why I thought it would be possible.

And what about failing a job inside a sandboxedworker?

it seems that u have added fail. But how do I have type annotion for that?

Just add fail into it?

[arthurvanl]

I use this with embedded mode. Is that a problem?

I also see this now:

But I am using upsertJobScheduler in embedded now. It does work, but do I need to switch?