Skip to content
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

Use extraFiles feature to cull idle kernels #563

Merged
merged 6 commits into from
Aug 27, 2021

Conversation

GeorgianaElena
Copy link
Member

@GeorgianaElena GeorgianaElena commented Jul 28, 2021

Todo:

  • Fix culling behavior
  • Docs (In pilot repo, for users) when culling happens (Document kernel culling docs#91)
  • Docs (In pilot-hubs repo, for us) where it is configured, and how to raise it

Fixes #518

@yuvipanda
Copy link
Member

yuvipanda commented Jul 29, 2021

This is great! Can you add a little docs too?

items moved above

Copy link
Contributor

@damianavila damianavila left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This one looks OK to me.
Wondering how this change will impact our users... I mean, do we need to ANN this somehow to be sure we are not breaking established workflows?

@choldgraf
Copy link
Member

choldgraf commented Aug 24, 2021

Good question - I guess the question is "what is the default behavior" for jupyterhub when this isn't configured? Because I assume that this what our behavior was, since this config wasn't taking effect right?

Maybe related to that - @GeorgianaElena could you add a short section in our docs maybe here: https://pilot.2i2c.org/en/latest/admin/howto/control-user-server.html that explains what this behavior is expected to be for users?

@GeorgianaElena
Copy link
Member Author

Good question - I guess the question is "what is the default behavior" for jupyterhub when this isn't configured? Because I assume that this what our behavior was, since this config wasn't taking effect right?

I think that before this, the kernels were just not being stopped, but rather the entire notebook servers after that default 1h.
But if this is true, I'm wondering whether we should have the kernel culling timeout be shorter than the one of notebook servers.

Right now, they are both 1h and I'm wondering what happens in this scenario, since the kernel stopping is seen by the notebook culler as "activity". Does this mean that the notebook server timeout is reset once a kernel is culled? Does this makes the notebook cull timeout be actually 1h + 1h? @yuvipanda, what do you think?

@damianavila
Copy link
Contributor

Right now, they are both 1h

Btw, I only see a server cull in the CB demo Hub: https://github.com/2i2c-org/pilot-hubs/blob/master/config/hubs/cloudbank.cluster.yaml#L245

Am I missing something or we do not have server culling in most of our Hubs?

@yuvipanda
Copy link
Member

@damianavila z2jh enables it by default https://github.com/jupyterhub/zero-to-jupyterhub-k8s/blob/8c4bed8c94fd14fc9292b9f9730a200404a5e6e8/jupyterhub/values.yaml#L554

Copy link
Member

@yuvipanda yuvipanda left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Copy link
Member

@choldgraf choldgraf left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This LGTM - I made one comment in the pilot/ docs but 👍 after that!

@damianavila
Copy link
Contributor

@damianavila z2jh enables it by default

And I was obviously missing something... I forgot to check the z2jh stuff before talking, thanks for the reminder!

@yuvipanda
Copy link
Member

Not sure if that was clear - but you can now merge this whenever you feel like, @GeorgianaElena

@GeorgianaElena GeorgianaElena merged commit 8b33364 into 2i2c-org:master Aug 27, 2021
@GeorgianaElena
Copy link
Member Author

Thanks everybody! 🌞

@choldgraf
Copy link
Member

Woohoo! 🚀

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

Fix in-notebook config for culling kernels
4 participants