-
Notifications
You must be signed in to change notification settings - Fork 835
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
EC2 Nodeclass using Bottlerocket image won't provision larger root volume #6214
Comments
I was able to reproduce, and it does look like Karpenter is provisioning the requested EBS volumes but that is not reflected in the root partition. I'm not a bottlerocket expert, but my understanding is that the root partition is immutable so it being a fixed size makes sense.
You can see that when I list the block devices from the admin container that the 5gb root volume is present.
What is your goal in increasing the size of the root partition? I do think this is likely more a bottlerocket question than a Karpenter question. |
Hi @jmdeal Thanks for looking into this. We were looking to increase the size of the root volume solely because it kept on triggering disk space low alerts in our monitoring systems. Since the Karpenter documentation mentioned that a custom disk size could be set, we thought that it would be an easy way to give the disk a bit more breathing room. If the disk size is actually immutable though, we will probably just silence the alerts. |
This issue has been inactive for 14 days. StaleBot will close this stale issue after 14 more days of inactivity. |
Description
Observed Behavior:
We are running Karpenter with AWS Bottlerocket nodes, but have seen that since moving to Bottlerocket AMIs from AL2 that the /dev/root volume is always >80% full.
The documentation states that a 4Gi EBS volume should be provisioned by default for /dev/root and a 20Gi EBS volume for data.
We are seeing, however, that only a 1Gi /dev/root volume is provisioned.
Doc link: https://karpenter.sh/v0.34/concepts/nodeclasses/#bottlerocket-1
We have therefore tried overriding the blockDeviceMapping both for root and data. This appears to work for the data volume, but not for root.
Checking mounted volume sizes on the instance one can see that /dev/nvme1n1p1 is at the expected 50Gi, however the /dev/root volume is still at 1Gi.
This only appears to be happening with Bottlerocket nodes as it is not affecting our AL2 nodes / EC2NodeClass.
Expected Behavior:
We are expecting that the root volume should be overridden to 5Gi and the data volume be overridden to 50Gi.
Reproduction Steps (Please include YAML):
Provisioning the EC2NodeClass:
Describing the EC2NodeClass after deploying the above config:
Checking the volume sizes on the instance shows that the /dev/root volume hasn't increased in size, while the "data" volume has increased as expected:
Versions:
kubectl version
): v1.29.3-eks-adc7111The text was updated successfully, but these errors were encountered: