[gaurang-delavadiya]

Description of the issue

#778 moved the AKS Container Insights suite onto the amazon-cloudwatch-observability Helm chart, but terraform/azure/aks still deploys a hand-written default:otel DaemonSet for the OTLP suite, and the suite is chosen with a test_mode variable that has to be kept in step with test_dir. AKS is also not in the test case generator, so the agent workflow carries one hand-written job per AKS suite, while EKS suites only need a generator entry.

Description of changes

• terraform/azure/aks installs the chart for both suites with helm_release (k8sMode=AKS). Both use the scripts/azure/setup.sh values from test(containerinsights): run OTel Container Insights suites through the Helm chart #778 (containerInsights.enabled=false, containerLogs.enabled=false). The Container Insights suite adds otelContainerInsights.enabled and otelContainerInsights.logs.enabled; the OTLP suite sets agent.config=default:otel with OTel Container Insights off.
• The image under test is set by patching the AmazonCloudWatchAgent resources, as the EKS stacks do. The ECR pull secret is attached to the cloudwatch-agent ServiceAccount and the agent workloads are restarted.
• The suite is selected from test_dir (validated). test_mode and the hand-written agent DaemonSet are removed.
• helm_charts_branch is renamed to helm_chart_branch, the name used by the EKS stacks and by the chart branch step in the agent workflow. helm_set_values allows extra chart values.
• The OTLP load generator uses the shared terraform/otlp_load_generator.sh with prefix = "aks".
• Generator: new aks test type with both suites, and generator/resources/aks_test_matrix.json (Standard_D4s_v7).
• The AKS Go tests are unchanged.

License

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.

Tests

• Both suites run end to end with terraform apply against AKS (eastus, Standard_D4s_v7), chart branch release-6.7.0 (the branch the workflow resolves today), agent image 7b574ef5:
  • ./test/azure/aks/containerinsights: TestAKSContainerInsights PASS (node, cluster, KEDA and Karpenter metrics; node logs).
  • ./test/azure/aks: TestAKS PASS (Metrics: OTLP, SpanMetrics, Names; Logs; Traces).
  • Both destroys completed.
• The GenerateTestMatrix and AKSIntegrationTest scripts from ci: run AKS integration tests from the generated test matrix amazon-cloudwatch-agent#2342 were run locally against this branch, with terraform plan in place of apply/destroy: aks_matrix for each test_dir_filter value, chart branch resolution, and plan / plan -destroy for both rows pass.
• go vet -tags integration ./test/azure/..., gofmt, terraform validate and terraform fmt -check pass.

Related pull requests

• Companion: ci: run AKS integration tests from the generated test matrix amazon-cloudwatch-agent#2342
• Merge order: this PR first, then ci: run AKS integration tests from the generated test matrix amazon-cloudwatch-agent#2342 straight after. The current AKS jobs pass test_mode, which this PR removes.