chore: Fix env vars used in e2e Karpenter install and diff test scripts

[bryantbiggs]

Fixes #N/A

Notice that the region is missing from the endpoint that is rendered in the script

Description

• Fix env vars used in e2e Karpenter install and diff test scripts
  • Since the URI format for ECR is using the ECR DKR endpoint, there shouldn't need to be a difference between endpoints for public cluster vs private cluster (unless the images are located in a different accounts). Therefore the endpoints should be the same between public and private cluster usage which means the same account ID and region. If my understanding of

    karpenter-provider-aws/.github/actions/e2e/install-karpenter/action.yaml

    Lines 51 to 67 in c925158

    	- name: install-karpenter
    	shell: bash
    	env:
    	ECR_ACCOUNT_ID: ${{ inputs.ecr_account_id }}
    	ECR_REGION: ${{ inputs.ecr_region }}
    	ACCOUNT_ID: ${{ inputs.account_id }}
    	CLUSTER_NAME: ${{ inputs.cluster_name }}
    	PRIVATE_CLUSTER: ${{ inputs.private_cluster }}
    	run: |
    	./test/hack/e2e_scripts/install_karpenter.sh
    	- name: diff-karpenter
    	shell: bash
    	env:
    	ECR_ACCOUNT_ID: ${{ inputs.ecr_account_id }}
    	ECR_REGION: ${{ inputs.ecr_region }}
    	run: |
    	./test/hack/e2e_scripts/diff_karpenter.sh

    is correct - ACCOUND_ID and REGION are the account ID and region where the clusters are created whereas the ECR_ACCOUNT_ID and ECR_REGION are the account ID and region where the Helm chart and images are stored within ECR. The prior scripts were switching between a mix of cluster account ID and region and the ECR account ID and region, which not all of those values are provided to the script and resulted in incorrect template rendering

How was this change tested?

Does this change impact docs?

• Yes, PR includes docs updates
• Yes, issue opened: #
• No

By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.

[netlify]

✅ Deploy Preview for karpenter-docs-prod canceled.

Name	Link
🔨 Latest commit	f175f32
🔍 Latest deploy log	https://app.netlify.com/sites/karpenter-docs-prod/deploys/66e431cddef9f50008a0ebf9

[bryantbiggs]

cc @jigisha620

[coveralls]

Pull Request Test Coverage Report for Build 10849227974

Details

• 0 of 0 changed or added relevant lines in 0 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage remained the same at 83.025%

---

Totals
Change from base Build 10842727395:	0.0%
Covered Lines:	5512
Relevant Lines:	6639

---

💛 - Coveralls

[jonathan-innis]

/karpenter snapshot

[github-actions]

Snapshot successfully published to oci://021119463062.dkr.ecr.us-east-1.amazonaws.com/karpenter/snapshot/karpenter:0-de4dda146397e7020060a40ea06067ba648f9070.
To install you must login to the ECR repo with an AWS account:

aws ecr get-login-password --region us-east-1 | docker login --username AWS --password-stdin 021119463062.dkr.ecr.us-east-1.amazonaws.com

helm upgrade --install karpenter oci://021119463062.dkr.ecr.us-east-1.amazonaws.com/karpenter/snapshot/karpenter --version "0-de4dda146397e7020060a40ea06067ba648f9070" --namespace "kube-system" --create-namespace \
  --set "settings.clusterName=${CLUSTER_NAME}" \
  --set "settings.interruptionQueue=${CLUSTER_NAME}" \
  --set controller.resources.requests.cpu=1 \
  --set controller.resources.requests.memory=1Gi \
  --set controller.resources.limits.cpu=1 \
  --set controller.resources.limits.memory=1Gi \
  --wait

[bryantbiggs]

/karpenter snapshot

[jonathan-innis]

@bryantbiggs Is this still relevant since the code was updated with a fix that resolved the failure that we were running into?

[bryantbiggs]

yes, its still relevant

there are a few parts:

1. Because the endpoint that is used is using the ECR private endpoint format, there shouldn't need to be a difference between the endpoint used for testing a private clusters versus a non-private cluster. The only time you would need to make decision on which endpoint would be if you were pulling images from a public registry like Dockerhub, Public ECR, etc. So the chart URI should be the same regardless of public or private cluster
2. It doesn't look like tests are currently executed against private clusters because not all of the required environment variables are being passed down to the scripts. For example this script needs the REGION environment variable but the workflow does not pass that environment variable. Likewise, this script requires both ACCOUNT_ID and REGION when its a private cluster, but those are not passed in the workflow
3. There seems to be a mixup between ACCOUNT_ID and ECR_ACCOUNT_ID as well as REGION and ECR_REGION - I can't say which is right or wrong without knowing the intent, but based on the two issues above - I think the ECR_* versions should be used whenever it comes to an OCI artifact coming out of ECR