Helm chart failing to install when IMDS is unavailable

[solmonk]

For the following configuration variables, we rely on IMDS to get the right values:

• REGION
• AWS_ACCOUNT_ID
• CLUSTER_VPC_ID
• CLUSTER_NAME

In many environments, IMDS is not available. (e.g. k8s w/o EC2 instance, self-managed clusters with IMDS access blocked) In this case, the values should be provided as environment variable through Helm chart. However, on values.yaml we have 3 of these but CLUSTER_NAME is missing.

This is causing any Helm chart installation to fail in IMDS-disabled environment. One case reported is kops which disables IMDS access for pods by default.

Path to fix:

• Add cluster name to helm chart value.
• Update documentation with proper IMDS setup instruction. Setting IMDS hops to 2 or higher is worth mentioning.

Temporary mitigation:

• If you are using IMDSv2, try increasing the hop limit to 2 or higher.
• Deploy manually. Modify deploy-v0.0.17.yaml and add above 4 environment variables.

[graehren]

Need IMDS related section in our documentation