Coiled inspect doesn't work in JupyterLab

[DPeterK]

If you try to inspect a coiled software environment in JupyterLab you hit an error:

coiled.inspect("tiledb")

---------------------------------------------------------------------------
RuntimeError                              Traceback (most recent call last)
<ipython-input-9-0829b31b39f9> in <module>
----> 1 coiled.inspect("tiledb")

/anaconda/envs/tiledb2/lib/python3.7/site-packages/coiled/cli/env.py in inspect(name)
     82 
     83     """
---> 84     results = get_event_loop().run_until_complete(get_software_info(name))
     85     for key in ["container", "conda", "pip", "post_build"]:
     86         print(f"{key}:")

/anaconda/envs/tiledb2/lib/python3.7/asyncio/base_events.py in run_until_complete(self, future)
    561         """
    562         self._check_closed()
--> 563         self._check_runnung()
    564 
    565         new_task = not futures.isfuture(future)

/anaconda/envs/tiledb2/lib/python3.7/asyncio/base_events.py in _check_runnung(self)
    521     def _check_runnung(self):
    522         if self.is_running():
--> 523             raise RuntimeError('This event loop is already running')
    524         if events._get_running_loop() is not None:
    525             raise RuntimeError(

RuntimeError: This event loop is already running

I get a near-identical error if I run coiled.install("tiledb").

I don't get either error if I just use a plain Python prompt. I'm using coiled v0.0.27 here.

Note this isn't preventing me from being able to use coiled at all. I just happened to come across it, so thought I'd raise 🙂

[jrbourbeau]

Thanks for letting us know about this @DPeterK! It looks like there's a poor interaction between the JupyterLab event loop and the event loop coiled is using. We'll look into it!

[mrocklin]

It looks like the inspect and maybe other CLI methods are creating their own event loop. In this case one thing we could do is take the functions in cloud/cli/utils.py and put them on the Cloud object. Maybe most logic/interactions should go on that object rather than in separate functions. CLI functions should maybe also be pointers to code in core.py rather than the other way around.

[jrbourbeau]

Thanks @mrocklin. I was just revisiting this issue as it came up elsewhere recently. I'll give moving some functions in cloud/cli/utils.py toCloud methods a shot.

[jrbourbeau]

Thanks for reporting @DPeterK! This will be fixed in the next coiled release

[DPeterK]

Great stuff! Thanks @jrbourbeau and @mrocklin