Bring your towel and join one of the fastest growing data communities. Welcome to our second-generation open source orchestration platform, a completely rethought approach to dataflow automation.

Prefect Community

Hi all, we are seeing the error "At least one upstream state has an unmappable result" for a failed task. A subsequent manual run of the same flow finishes without error.  We think this might only happen when the upstream task has to retry several times before succeeding, but I've failed to reproduce this. Any idea what might be causing this?

This error should occur if any input to a mapped task is unmappable. For example:

```task.map([1, 2, 3], 1)  # the 2nd parameter isn't mappable```
If all inputs to mapped tasks should be mappable, then there might be a bug (or something else going on). We'd need a reproducible example to help debug further.

Ok I think I have a minimal reproduction:

`import datetime`
`import prefect`

`@prefect.task(max_retries=10, retry_delay=datetime.timedelta(seconds=900))`
`def task_a():`
    `if prefect.context.task_run_count &lt; 2:`
        `raise Exception()`
    `return 'a result'`

`@prefect.task`
`def task_b(a, c):`
    `return 'b result'`

`with prefect.Flow('flow name') as flow:`
    `a = task_a()`
    `b = task_b.map(prefect.unmapped(a), [1, 2, 3])`

`flow.run_config = prefect.run_configs.DockerRun()`

`flow.storage = prefect.storage.Docker(`
    `# python image`
`)`

`flow.register('project')`

I've only done pretty limited testing, but it seems important that the waiting time is long (somewhere at or above 10-15 minutes) and that it uses docker storage.

There is a separate code path for retries &gt; 10 min (the flow run should stop and be resubmitted to pick up again later) so that makes sense. I'll open an issue from this thread, thanks.

<@ULVA73B9P> open "Mapped tasks fail if an upstream task has a retry &gt; 10 min"

<https://github.com/PrefectHQ/prefect/issues/3990>

Thanks, Jim. I'll keep an eye on this issue.