Failure in ProfilesSpawner.load_child_class() with empty/no sqlite database

[rcthomas]

This is with JuptyterHub 0.8.0b5 --- when I start up without any pre-existing sqlite database I still get the KeyError. Maybe a recent change there causes this to fail now?

[SaintSimmo]

Same issue. Got around it by defining a default child_profile that exists (where before it was simple ''). Now stuck on #11 which seems to be a bigger issue.

[jrdemasi]

@SaintSimmo Can you elaborate on how you got around it for users who are less familiar with what you're talking about?

[SaintSimmo]

@jrdemasi I ultimately ended up going with imagespawner (https://github.com/ryanlovett/imagespawner) for choosing a docker image upon starting up one's personal server. If you're using docker images, this is what I'd go with for right now at least.

I never ended up being able to get wrapspawner to fully work for my needs.

[opoplawski]

I'm seeing this as well with jupyterhub 0.8.1.

[jrdemasi]

I ended up just downgrading for now. No time to mess with it.

…

On Nov 8, 2017 11:39, "Orion Poplawski" ***@***.***> wrote: I'm seeing this as well with jupyterhub 0.8.1. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#10 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ARqQJ1dCrbHj9Lh9cT0QtUI6B8TeOYeuks5s0fVngaJpZM4Pf62Y> .

[rbhatia5]

I'm also seeing this issue, curious if anyone's solved it well

[chiroptical]

Is this the issue you all are seeing? I don't want to submit another issue if it is.

Using:

• Python 3.5.2 :: Anaconda 4.2.0 (64-bit)
• JupyterHub: v0.8.1 (conda install -c conda-forge jupyterhub=0.8.1)
• {Wrap,Batch}Spawner most recent git master branch

Issue:

HTTPServerRequest(protocol='http', host='<redacted>', method='GET', uri='/hub/home', version='HTTP/1.1', remote_ip='<redacted>', headers=<redacted>)
Traceback (most recent call last):
    File "<path>/lib/python3.5/site-packages/tornado/web.py", line 1469, in _execute
    result = yield result
    File "<path>/lib/python3.5/site-packages/jupyterhub/handlers/pages.py", line 67, in get
    if user.running:
    File "<path>/lib/python3.5/site-packages/jupyterhub/user.py", line 240, in running
    return self.spawner.ready
    File "<path>/lib/python3.5/site-packages/jupyterhub/user.py", line 215, in spawner
    return self.spawners['']
    File "<path>/lib/python3.5/site-packages/jupyterhub/user.py", line 105, in __getitem__
    self[key] = self.spawner_factory(key)
    File "<path>/lib/python3.5/site-packages/jupyterhub/user.py", line 209, in _new_spawner
    spawner.load_state(orm_spawner.state or {})
    File "<path>/lib/python3.5/site-packages/wrapspawner/wrapspawner.py", line 91, in load_state
    self.load_child_class(state)
    File "<path>/lib/python3.5/site-packages/wrapspawner/wrapspawner.py", line 211, in load_child_class
    raise KeyError('jupyterhub database might be outdated, please reset it, in the default configuration, just delete jupyterhub.sqlite')
KeyError: 'jupyterhub database might be outdated, please reset it, in the default configuration, just delete jupyterhub.sqlite'

[sknigh]

I am getting the same error output as @barrymoo from a virtual environment-based installation. Are there any workarounds for this?

[jrdemasi]

I ended up downgrading as opposed to wasting time on it. Unless you need a specific feature in 8.x, I think that's the easiest course of action.

…

On Dec 7, 2017 12:16 PM, "sknigh" ***@***.***> wrote: I am getting the same error output as @barrymoo <https://github.com/barrymoo> from a virtual environment-based installation. Are there any workarounds for this? — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#10 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ARqQJ5e6KCGxm4xv1N8veDAKRGI3_ur3ks5s-Dl7gaJpZM4Pf62Y> .

[sknigh]

Unfortunately I need some of the bug fixes in 0.8.

Sticking this to the end of my installation script makes profilespawner work with my slurm queues.

sed -i '75i\                oauth_client_id = self.oauth_client_id,' /usr/lib/python3.4/site-packages/wrapspawner/wrapspawner.py
sed -i '75i\                server = self._server,' /usr/lib/python3.4/site-packages/wrapspawner/wrapspawner.py
sed -i "213s/.*/            self.child_profile = ''/" /usr/lib/python3.4/site-packages/wrapspawner/wrapspawner.py

[Luke035]

Same issue here with Jupyterhub 0.8.1, Python 3.5.4 Anaconda 5.0.1.
My profiles are simple BatchSpawner profiles that I've always run with Jupyterhub 0.7.2

[rcthomas]

Have a look here.

master...NERSC:non-fatal-if-no-db

I never submitted it as a PR as I wasn't able to work through what all this change impacted and I wasn't able to fully test it due to #11 but thought I'd mention it. Haven't had time to get back to either issue yet...

[sknigh]

@rcthomas Oauth doesn't work because jupyterhub 0.8 configures it for the set spawner, but wrapspawner hasn't been updated to copy that information to it's child profile. See the first sed insertion from my comment above.

[rcthomas]

@sknigh I tried your fix and it works for me on 0.8 --- any reason you haven't submitted that as a PR addressing this issue and #11 both?