-
Notifications
You must be signed in to change notification settings - Fork 24
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
feat: Update EFA device plugin to use the EKS charts repo #26
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@vara-bonthu to validate
repository = try(var.aws_efa_k8s_device_plugin_helm_config["repository"], null) | ||
chart = try(var.aws_efa_k8s_device_plugin_helm_config["chart"], "${path.module}/helm-charts/aws-efa-k8s-device-plugin") | ||
version = try(var.aws_efa_k8s_device_plugin_helm_config["version"], "0.1.0") | ||
repository = try(var.aws_efa_k8s_device_plugin_helm_config["repository"], "https://aws.github.io/eks-charts") |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I like to use the upstream repo instead of our local one, have you validated that all the configs that we are using by default in the values.yaml
file are used into the default chart values? I'm asking that because people could be relying in the default values. @vara-bonthu need you on that.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thats a good call, I haven't checked what those defaults are in the local copy. I'll double check those and make sure we aren't breaking anything.
We may have added tolerations that the upstream doesn't.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looking at a diff of the values:
- Upstream Drops
beta.kubernetes.io/instance-type
from the label selector (Starting in v1.17, this label is deprecated in favor of node.kubernetes.io/instance-type.) - Upstream moves version from
v0.3.3
tov0.4.4
- upstream has added a requests block to the values:
resources:
requests:
cpu: 10m
memory: 20Mi
- our Local copy has a toleration to run everywhere, upstream values has an empty toleration config.
tolerations:
- operator: Exists # DaemonSet is tolerant of any taints, regardless of the key or value of the taint.
- key: CriticalAddonsOnly
operator: Exists
I'll add that toleration to our addon config but I think the other changes are ok.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I tried to follow the way the Nvidia plugin configures those values
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
What does this PR do?
Updates aws-efa-k8s-device-plugin.tf to use the eks-charts repo (https://github.com/aws/eks-charts/tree/master/stable/aws-efa-k8s-device-plugin) instead of a local copy of the device plugin chart
Motivation
this should make it easier to update later and includes the latest updates and instance types in the nodeAffinity.
More
pre-commit run -a
with this PRFor Moderators
Additional Notes