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

Suggestion: Kata deployment scripts should check that nested virtualization is enabled #89

Open
olberger opened this issue Apr 20, 2019 · 8 comments
Labels
bug Something isn't working

Comments

@olberger
Copy link

As exhibited in #87 crio install won't work whenever no nested virtualization has been allowed for KVM on the host.

I think it would be great to have some check when qemu/kvm will be used for kata-containers, to abort or warn prominently if no nested vrtualization is available, which will make qemu to fail, thus crio too.

Thanks in advance.

@krsna1729
Copy link
Contributor

@olberger since the script itself runs only on master, maybe updating https://github.com/kata-containers/packaging/tree/master/kata-deploy is a better idea?

In this repo we could add all examples to make use of node-selector with kata-runtime true label. That way they wont land on nodes that do not support kata-runtime class. This is a workaround for the current limitation in runtimeclass.

cc @egernst @mcastelino

@krsna1729
Copy link
Contributor

@olberger thanks for the suggestion!

@mcastelino
Copy link
Contributor

@egernst does your kata-deploy script check for this today?

@egernst
Copy link
Contributor

egernst commented Apr 22, 2019

no

@mcastelino
Copy link
Contributor

We should see if we can use NFD to detect support for VT-x and use that as part of kata-deploy when available.

/cc @krsna1729

@krsna1729
Copy link
Contributor

@mcastelino that will only save a daemonset pod. It doesnt solve the problem of scheduling the actual workload itself

@mythi
Copy link
Contributor

mythi commented Jun 20, 2019

that will only save a daemonset pod. It doesnt solve the problem of scheduling the actual workload itself

True but AFAICS the limitation comes from RuntimeClass that assumes homogeneous cluster.

@krsna1729
Copy link
Contributor

Yes see second comment

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
bug Something isn't working
Projects
None yet
Development

No branches or pull requests

6 participants