examples/

Tests layout and local/cluster mode

How it works

• You run your tests in plain go test locally
• You define cluster entrypoint in zcluster and use the same go test but with a cluster entrypoint, it builds and pushes images to ECR and run tests with Helm
• When each job got allocated your pod code will start to wait until all pods spun by client ( with unique sync id generated on start ) will have status Running if that won't happen until timeout client will remove all the pods
• When N pods with unique test label are in status Running the test will start, lag between pods start is 1sec max
• In case of any fatal error client will remove all the jobs
• When all jobs are complete test will end

Prepare k8s namespace

• Create namespace wasp and apply default permissions

cd charts/wasp
kubectl create ns wasp
kubectl -n wasp apply -f setup.yaml

Layout

In order to build the tests properly your test directory layout should follow simple rules:

• It should have go.mod
• It should be buildable with go test -c ./... from a dir where go.mod exists

Overall layout looks like:

tests
├── group1
│   ├── test_case_1_test.go
│   ├── gun.go
│   └── vu.go
├── group2
│   ├── subgroup21
│   │   ├── test_case_2_test.go
│   │   ├── gun.go
│   │   └── vu.go
└── go.mod

Build

Default Dockerfile and build_script are working with AWS ECR private repos and building only for amd64 platform

These vars are required to rebuild default layout and update an image

		UpdateImage:  true,
		BuildCtxPath: "..",
		
		// Helm values
		"image": ...

Turn off UpdateImage if you want to skip the build

If for some reason you don't like this layout or can't build like go test -c ./..., or you would like to customize your builds then you need to customize default Dockerfile and build_script and reference them in cluster_entrypoint

		DockerfilePath: "",
		BuildScriptPath: "",

Deployment

Default helm chart is here

If no chart override is provided we are using this tar

You can set your custom chart as a local path or as an OCI registry URI in cluster_entrypoint

• Set wasp chart in test params

ChartPath: "../../charts/wasp"
or 
ChartPath: "oci://public.ecr.aws/chainlink/wasp"

Deployment values

• This Helm values are usually static or secret, use any configuration to set them, simplest is just as env vars

                        // secrets
			"env.loki.url":        os.Getenv("LOKI_URL"),
			"env.loki.token":      os.Getenv("LOKI_TOKEN"),
			"env.loki.basic_auth": os.Getenv("LOKI_BASIC_AUTH"),
			"env.loki.tenant_id":  os.Getenv("LOKI_TENANT_ID"),
			
			// WASP vars
			"env.wasp.log_level": "debug",

			
			// test vars
			"image":               os.Getenv("WASP_TEST_IMAGE"),
			"test.binaryName":     os.Getenv("WASP_TEST_BIN"),
			"test.name":           os.Getenv("WASP_TEST_NAME"),
			"test.timeout":       "24h",
			"jobs":               "40",
			
			// k8s pods resources
			"resources.requests.cpu":    "2000m",
			"resources.requests.memory": "512Mi",
			"resources.limits.cpu":      "2000m",
			"resources.limits.memory":   "512Mi",
			
			// your custom vars that will pass through to k8s jobs
			"test.MY_CUSTOM_VAR": "abc",

Code examples

Running tests

Use go test for local runs

go test -v -run TestNodeRPS

Use go test for remote runs (consume vars from any config method, example is env vars)

export WASP_TEST_TIMEOUT="12h"
export WASP_TEST_IMAGE="..."
export WASP_TEST_BIN="profiles.test"
export WASP_TEST_NAME="TestNodeRPS"
export WASP_UPDATE_IMAGE="true"
go test -v -run TestClusterEntrypoint