-
Notifications
You must be signed in to change notification settings - Fork 904
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
Karpenter is nominating machine that have not enough memory #4450
Comments
This bug should be fixed from kubernetes-sigs/karpenter#449 and will be released early next week. |
@bwagner5 it looks like there is still now new release... we are also suffering from this issue. |
Tested with 0.30.0-rc.0. The same behaviour. I mean Karpenter provisions new nodes (in this case c5d.large), but the pending pod cannot be scheduled there. Just I remove one of the Daemonsets (e.g. node-problem-detector) then the pending pod can be scheduled.
If I compare the memory resources then I can see difference between the Karpenter calculation and the allocatable resources:
|
In 0.30.0 the same problem exists @bwagner5 @jonathan-innis |
It looks like this has to do with the One workaround for this is to raise your |
So following this it just seems that if we are using other OS (Bottlerocket) there will be difference in the allocatable of the nodes and the list in #4517 will not apply for Bottlerocket and it will need to be regenerated specifically. |
From the looks of the generated data, it looks like you may need to put it up to
We are planning to gen a list that considers all of our supported OSs and then either:
|
Closing in favor of #5161 |
Description
Observed Behavior:
The pod is still in Pending state, Karpenter is nominating a mchine that have not enough memory.
The text was updated successfully, but these errors were encountered: