Broadcast: GraphQL timeout error for many namespaces

[ColemanTom]

Description

I'm not sure if this is related to #6222, but, via the WUI, I did a broadcast to many namespaces and got this time out error from GraphQL.

2024-08-28T06:00:29Z INFO - Command "broadcast" received from USERNAME.
    broadcast(cycle_points=['*'], mode=put_broadcast, namespaces=['task_000', 'task_012', 'task_024', 'task_048', 'task_060', 'task_072', 'task_084', 'task_096', 'task_108', 'task_120', 'task_132', 'task_144', 'task_156', 'task_168', 'task_180', 'task_192', 'task_204', 'task_216', 'task_228', 'task_240'], settings=[{'execution time limt': 'PT30M'}])
Traceback (most recent call last):
  File "miniconda3/cylc-8.3.3/lib/python3.11/site-packages/promise/promise.py", line 844, in handle_future_result
    resolve(future.result())
            ^^^^^^^^^^^^^^^
  File "miniconda3/cylc-8.3.3/lib/python3.11/site-packages/cylc/flow/network/graphql.py", line 433, in async_resolve
    return await next_(root, info, **args)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "miniconda3/cylc-8.3.3/lib/python3.11/site-packages/promise/iterate_promise.py", line 10, in iterate_promise
    yield from promise.future  # type: ignore
    ^^^^^^^^^^^^^^^^^^^^^^^^^
graphql.error.located_error.GraphQLLocatedError: execution time limt

Reproducible Example

Will see if I can provide a simple one, but none provided yet. Just broadcast to many tasks at once. I've not been able to replicate it from CLI.

Expected Behaviour

No error, broadcasts get through. Maybe you need to split the broadcast up into chunks of N size?

[oliver-sanders]

@ColemanTom, I haven't been able to replicate this one. Where did you observe this traceback, in the scheduler log or the Cylc UI Server log?

It's quite likely that this is related to #6334 given the execution time limit key. Might be worth trying to reproduce it with the fix for that issue if possible.

[ColemanTom]

Cylc log. I don't even know where the ui log is to look at that.

[oliver-sanders]

~/.cylc/uiserver/log

[ColemanTom]

It could simply be a slow network caused it. I was running things through the WUI for another user too if that adds any latency, it wasn't as myself. I'll see if I can do it again as myself and as a different user again later or next week.

[oliver-sanders]

Aah, I've just spotted what's going on here. It's a typo limt rather than "limit".

graphql.error.located_error.GraphQLLocatedError: execution time limt

It is erroring because the config broadcasted is invalid, however, the error isn't being caught and logged nicely.

Have opened an issue to cover this: #6429