New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Have the JupyterHub spawner report issues with spawning the user's server #505
Comments
I believe errors creating the pod will result in K8s events for the pod. For example, if there is a problem pulling the image or no node can schedule the K8s pod; this should generate K8s events. I'm not sure off hand how we'd surface those in the UI; but a minimum we should be able to provide kubectl get commands and other instructions to help troubleshooting. |
/assign @pdmack |
@pdmack Are you making progress? Do you need help? |
I've asked on gitter when we might expect another kubespawner release. |
Sounds like we will get this simultaneously with the official JH 0.9 release. @willingc ? |
@pdmack yeah, I think it'll come out with JH 0.9 - or a little bit earlier. |
#905 upgrade JupyterHub and KubeSpawner. The new UI is much better. There is a status bar and it looks like it shows K8s events related to the pod as well. I'm going to mark this as fixed with the expectation that we will file new issues related to the latest version of JupyterHub and KubeSpawner as needed. |
Very glad to see this getting into kubeflow! Special thanks to @minrk for the progress api implementation in jupyterhub as well as reviewing my initial PR. |
Right now, if you use the JupyterHub spawner to spawn a server, and the corresponding notebook server fails to start, then you can wind up just seeing a 500-error page with no details (after which, you have to resort to running
kubectl get ...
to debug the issue).I think a lot of potential user confusion/frustration could be avoided if we could report the underlying issue on the error page instead of only showing a 500-error.
For example, the first time I tried using Kubeflow, I got an issue where I had accidentally undersized my cluster, so my server failed to start because the pod couldn't be scheduled. If the issue was made clear by the JupyterHub UI, then I would have resolved the issue much faster than I did.
I would like to report at least the following potential sources of errors if they occur:
There may be some parts of this that warrant upstream contributions to JupyterHub, but we should still have at tracking issue in Kubeflow for it, since the KubeFormSpawner would need to incorporate whatever eventually gets built for this.
The text was updated successfully, but these errors were encountered: