e2e: existing_arrikto: initial e2e test (kubeflow#4154)

* kfctl: existing_arrikto: initial e2e test Signed-off-by: Yannis Zarkadas <yanniszark@arrikto.com> * fix deployment name and review comment Signed-off-by: Yannis Zarkadas <yanniszark@arrikto.com> * fix bad syntax with extend Signed-off-by: Yannis Zarkadas <yanniszark@arrikto.com> * increase timeout for service ip discovery Signed-off-by: Yannis Zarkadas <yanniszark@arrikto.com> * update deployments to reflect latest manifests Signed-off-by: Yannis Zarkadas <yanniszark@arrikto.com> * skip knative for existing_arrikto Signed-off-by: Yannis Zarkadas <yanniszark@arrikto.com> * bump machine size and increase timeout Signed-off-by: Yannis Zarkadas <yanniszark@arrikto.com> * python script: fix error in delete step Related: googleapis/google-api-python-client#299 Signed-off-by: Yannis Zarkadas <yanniszark@arrikto.com> * be more conservative in adding new test Signed-off-by: Yannis Zarkadas <yanniszark@arrikto.com> * revert checklist Signed-off-by: Yannis Zarkadas <yanniszark@arrikto.com>
StatCan · Oct 14, 2019 · 269f5c9 · 269f5c9
1 parent 3b27111
commit 269f5c9
Show file tree

Hide file tree

Showing 12 changed files with 188 additions and 33 deletions.
diff --git a/bootstrap/pkg/kfapp/existing_arrikto/existing.go b/bootstrap/pkg/kfapp/existing_arrikto/existing.go
@@ -411,7 +411,7 @@ func generateCert(addr string) ([]byte, []byte, error) {
 
 func getLBAddress(kubeclient client.Client) (string, error) {
 	// Get IngressGateway Service's address
-	const maxRetries = 40
+	const maxRetries = 80
 	var lbIngresses []corev1.LoadBalancerIngress
 	svc := &corev1.Service{}
 	lbServiceName := types.NamespacedName{Name: "istio-ingressgateway", Namespace: "istio-system"}

diff --git a/prow_config.yaml b/prow_config.yaml
@@ -85,6 +85,30 @@ workflows:
     kwargs:
       use_basic_auth: true
       config_path: https://raw.githubusercontent.com/kubeflow/manifests/master/kfdef/kfctl_gcp_basic_auth.yaml
+  # E2E tests for kfctl_existing_arrikto 
+  - app_dir: kubeflow/kubeflow/testing/workflows
+    component: kfctl_go_test
+    name: kfctl-go-existing
+    job_types:
+      # Enable once we have confirmed the stability of the test
+      # - presubmit
+      - postsubmit
+    include_dirs:
+      - bootstrap/*
+      - dependencies/*
+      - kubeflow/*
+      - testing/*
+    params:
+      platform: gke
+      gkeApiVersion: v1
+      workflowName: kfctl-go
+      useBasicAuth: false
+      useIstio: true
+      testEndpoint: false
+      configPath: https://raw.githubusercontent.com/kubeflow/manifests/master/kfdef/kfctl_existing_arrikto.yaml
+      cluster_creation_script: create_existing_cluster.sh
+      cluster_deletion_script: delete_existing_cluster.py
+      nameSuffix: existing_arrikto
   # Only run  kfctl presubmit test with basic auth if
   # files related to basic auth are modified.
   - py_func: kubeflow.kubeflow.ci.kfctl_e2e_workflow.create_workflow

diff --git a/testing/kfctl/conftest.py b/testing/kfctl/conftest.py
@@ -38,6 +38,14 @@ def pytest_addoption(parser):
       "--use_istio", action="store", default="False",
       help="Use istio.")
 
+  parser.addoption(
+      "--cluster_creation_script", action="store", default="",
+      help="The script to use to create a K8s cluster before running kfctl.")
+
+  parser.addoption(
+      "--cluster_deletion_script", action="store", default="",
+      help="The script to use to delete a K8s cluster before running kfctl.")
+
 @pytest.fixture
 def app_path(request):
   return request.config.getoption("--app_path")
@@ -62,6 +70,14 @@ def project(request):
 def config_path(request):
   return request.config.getoption("--config_path")
 
+@pytest.fixture
+def cluster_creation_script(request):
+  return request.config.getoption("--cluster_creation_script")
+
+@pytest.fixture
+def cluster_deletion_script(request):
+  return request.config.getoption("--cluster_deletion_script")
+
 @pytest.fixture
 def build_and_apply(request):
   value = request.config.getoption("--build_and_apply").lower()

diff --git a/testing/kfctl/kf_is_ready_test.py b/testing/kfctl/kf_is_ready_test.py
@@ -4,14 +4,15 @@
 import subprocess
 import tempfile
 import uuid
+import yaml
 from retrying import retry
 
 import pytest
 
 from kubeflow.testing import util
 from testing import deploy_utils
 
-def test_kf_is_ready(namespace, use_basic_auth, use_istio):
+def test_kf_is_ready(namespace, use_basic_auth, use_istio, app_path):
   """Test that Kubeflow was successfully deployed.
 
   Args:
@@ -32,37 +33,67 @@ def test_kf_is_ready(namespace, use_basic_auth, use_istio):
   # Verify that components are actually deployed.
   # TODO(jlewi): We need to parameterize this list based on whether
   # we are using IAP or basic auth.
+  # TODO(yanniszark): This list is incomplete and missing a lot of components.
   deployment_names = [
       "argo-ui",
       "centraldashboard",
-      "cloud-endpoints-controller",
       "jupyter-web-app-deployment",
-      "metadata-db",
-      "metadata-deployment",
-      "metadata-ui",
+      "minio",
       "ml-pipeline",
+      "ml-pipeline-persistenceagent",
       "ml-pipeline-scheduledworkflow",
       "ml-pipeline-ui",
+      "ml-pipeline-viewer-controller-deployment",
+      "mysql",
       "notebook-controller-deployment",
-      "tf-job-operator",
+      "profiles-deployment",
       "pytorch-operator",
-      "katib-controller",
+      "tf-job-operator",
       "workflow-controller",
   ]
 
-  stateful_set_names = [
-    "kfserving-controller-manager",
+  stateful_set_names = []
+
+  with open(os.path.join(app_path, "app.yaml")) as f:
+    kfdef = yaml.safe_load(f)
+  platform = kfdef["spec"]["platform"]
+
+  ingress_related_deployments = [
+    "istio-citadel",
+    "istio-egressgateway",
+    "istio-galley",
+    "istio-ingressgateway",
+    "istio-pilot",
+    "istio-policy",
+    "istio-sidecar-injector",
+    "istio-telemetry",
+    "istio-tracing",
+    "kiali",
+    "prometheus",
   ]
-
-  ingress_related_deployments = []
   ingress_related_stateful_sets = []
 
-  if use_basic_auth:
-    deployment_names.extend(["basic-auth-login"])
-    ingress_related_stateful_sets.extend(["backend-updater"])
-  else:
-    ingress_related_deployments.extend(["iap-enabler"])
-    ingress_related_stateful_sets.extend(["backend-updater"])
+  knative_namespace = "knative-serving"
+  knative_related_deployments = [
+          "activator",
+          "autoscaler",
+          "controller",
+  ]
+
+  if platform == "gcp":
+    deployment_names.extend(["cloud-endpoints-controller"])
+    stateful_set_names.extend(["kfserving-controller-manager"])
+    if use_basic_auth:
+      deployment_names.extend(["basic-auth-login"])
+      ingress_related_stateful_sets.extend(["backend-updater"])
+    else:
+      ingress_related_deployments.extend(["iap-enabler"])
+      ingress_related_stateful_sets.extend(["backend-updater"])
+  elif platform == "existing_arrikto":
+    deployment_names.extend(["dex"])
+    ingress_related_deployments.extend(["authservice"])
+    knative_related_deployments = []
+
 
   # TODO(jlewi): Might want to parallelize this.
   for deployment_name in deployment_names:
@@ -89,15 +120,9 @@ def test_kf_is_ready(namespace, use_basic_auth, use_istio):
 
   # TODO(jlewi): We should verify that the ingress is created and healthy.
 
-  knative_namespace = "knative-serving"
-  knative_related_deployments = [
-          "activator",
-          "autoscaler",
-          "controller",
-  ]
   for deployment_name in knative_related_deployments:
-      logging.info("Verifying that deployment %s started...", deployment_name)
-      util.wait_for_deployment(api_client, knative_namespace, deployment_name, 10)
+    logging.info("Verifying that deployment %s started...", deployment_name)
+    util.wait_for_deployment(api_client, knative_namespace, deployment_name, 10)
 
 if __name__ == "__main__":
   logging.basicConfig(level=logging.INFO,

diff --git a/testing/kfctl/kfctl_delete_test.py b/testing/kfctl/kfctl_delete_test.py
@@ -37,7 +37,14 @@ def get_endpoints_list(project):
 
   return endpoints
 
-def test_kfctl_delete(kfctl_path, app_path, project):
+def test_kfctl_delete(kfctl_path, app_path, project, cluster_deletion_script):
+
+  # TODO(yanniszark): split this into a separate workflow step
+  if cluster_deletion_script:
+    logging.info("cluster_deletion_script specified: %s", cluster_deletion_script)
+    util.run(["/bin/bash", "-c", cluster_deletion_script])
+    return
+
   if not kfctl_path:
     raise ValueError("kfctl_path is required")
 

diff --git a/testing/kfctl/kfctl_go_test.py b/testing/kfctl/kfctl_go_test.py
@@ -6,7 +6,7 @@
 from kubeflow.kubeflow.ci import kfctl_go_test_utils as kfctl_util
 from kubeflow.testing import util
 
-def test_build_kfctl_go(app_path, project, use_basic_auth, use_istio, config_path, build_and_apply):
+def test_build_kfctl_go(app_path, project, use_basic_auth, use_istio, config_path, build_and_apply, cluster_creation_script):
   """Test building and deploying Kubeflow.
 
   Args:
@@ -15,6 +15,7 @@ def test_build_kfctl_go(app_path, project, use_basic_auth, use_istio, config_pat
     use_basic_auth: Whether to use basic_auth.
     use_istio: Whether to use Istio or not
     config_path: Path to the KFDef spec file.
+    cluster_creation_script: script invoked to create a new cluster
     build_and_apply: whether to build and apply or apply
   """
   # Need to activate account for scopes.
@@ -24,11 +25,18 @@ def test_build_kfctl_go(app_path, project, use_basic_auth, use_istio, config_pat
         "--key-file=" + os.environ["GOOGLE_APPLICATION_CREDENTIALS"]
     ])
 
+  # TODO(yanniszark): split this into a separate workflow step
+  if cluster_creation_script:
+      logging.info("Cluster creation script specified: %s", cluster_creation_script)
+      util.run(["/bin/bash", "-c", cluster_creation_script])
+
+
   kfctl_path = kfctl_util.build_kfctl_go()
   app_path = kfctl_util.kfctl_deploy_kubeflow(
                   app_path, project, use_basic_auth,
                   use_istio, config_path, kfctl_path, build_and_apply)
-  kfctl_util.verify_kubeconfig(app_path)
+  if not cluster_creation_script:
+      kfctl_util.verify_kubeconfig(app_path)
 
 if __name__ == "__main__":
   logging.basicConfig(

diff --git a/testing/kfctl/scripts/create_existing_cluster.sh b/testing/kfctl/scripts/create_existing_cluster.sh
@@ -0,0 +1,34 @@
+#!/bin/bash
+
+set -e
+
+export PROJECT="kubeflow-ci"
+export GCP_ZONE="us-central1-a"
+export GCP_USER="$(gcloud config list account --format "value(core.account)" )"
+export GCP_PROJECT="$(gcloud config list project --format "value(core.project)" )"
+export CLUSTER_NAME="kfctl-arr-${REPO_NAME}-${BUILD_ID}"
+export CLUSTER_VERSION="$(gcloud container get-server-config --zone=${GCP_ZONE} --format="value(validMasterVersions[0])" )"
+
+############################
+# Create and setup cluster #
+############################
+
+gcloud container clusters create "${CLUSTER_NAME}" \
+--project "${GCP_PROJECT}" \
+--zone "${GCP_ZONE}" \
+--username "admin" \
+--cluster-version "${CLUSTER_VERSION}" \
+--machine-type "custom-6-23040" --num-nodes "1" \
+--image-type "UBUNTU" \
+--local-ssd-count=4 \
+--disk-type "pd-ssd" --disk-size "50" \
+--no-enable-cloud-logging --no-enable-cloud-monitoring \
+--no-enable-ip-alias \
+--enable-network-policy \
+--enable-autoupgrade --enable-autorepair
+
+echo "Getting credentials for newly created cluster..."
+gcloud container clusters get-credentials "${CLUSTER_NAME}" --zone="${GCP_ZONE}"
+
+echo "Setting up GKE RBAC..."
+kubectl create clusterrolebinding cluster-admin-binding --clusterrole=cluster-admin --user="${GCP_USER}"
diff --git a/testing/kfctl/scripts/delete_existing_cluster.py b/testing/kfctl/scripts/delete_existing_cluster.py
@@ -0,0 +1,28 @@
+#!/usr/bin/env python3
+
+import os
+import logging
+from googleapiclient import discovery
+from oauth2client.client import GoogleCredentials
+from kubeflow.testing import util
+
+
+def must_getenv(name):
+    value = os.getenv(name)
+    if not name:
+        logging.fatal("Environment variable %s is not set", name)
+        raise ValueError()
+    return value
+
+
+if __name__ == "__main__":
+
+    util.run([
+        "gcloud", "auth", "activate-service-account", "--key-file",
+        must_getenv("GOOGLE_APPLICATION_CREDENTIALS")
+    ])
+
+    cluster_name = "kfctl-arr-" + must_getenv("REPO_NAME") + "-" + must_getenv("BUILD_ID")
+    credentials = GoogleCredentials.get_application_default()
+    service = discovery.build('container', 'v1', credentials=credentials, cache_discovery=False)
+    util.delete_cluster(service, cluster_name, "kubeflow-ci", "us-central1-a")
diff --git a/testing/test_deploy.py b/testing/test_deploy.py
@@ -241,7 +241,7 @@ def test_successful_deployment(deployment_name):
     if i == retries:
       raise Exception('Deployment failed: ' + deployment_name)
     try:
-      output = util.run(["kubectl", "get", "deployment", deployment_name])
+      output = util.run(["kubectl", "get", "deployment", deployment_name, "-n", "kubeflow"])
       logging.info("output = \n" + output)
       if output.count('\n') == 1:
         output = output.split('\n')[1]

diff --git a/testing/workflows/components/kfctl_go_test.jsonnet b/testing/workflows/components/kfctl_go_test.jsonnet
@@ -33,7 +33,13 @@ local runPath = srcDir + "/testing/workflows/run.sh";
 local kfCtlPath = srcDir + "/bootstrap/bin/kfctl";
 local kubeConfig = testDir + "/kfctl_test/.kube/kubeconfig";
 
-// Name for the Kubeflow app.
+// cluster_creation_script specifies the script to run in order to create 
+// a cluster before running kfctl.
+// Only applicable to configs that don't create their own clusters.
+local cluster_creation_script = if (params.cluster_creation_script=="") then "" else srcDir + "/testing/kfctl/scripts/" + params.cluster_creation_script;
+local cluster_deletion_script = if (params.cluster_deletion_script=="") then "" else srcDir + "/testing/kfctl/scripts/" + params.cluster_deletion_script;
+
+
 // This needs to be unique for each test run because it is
 // used to name GCP resources
 // We take the suffix of the name because it should provide some random salt.
@@ -185,11 +191,13 @@ local nameSuffix1 = if util.toBool(params.useBasicAuth) then
   "basic-auth"
   else
   "iap";
-local nameSuffix = if util.toBool(params.useIstio) then
+local nameSuffix2 = if util.toBool(params.useIstio) then
   nameSuffix1 + "-istio"
   else
   nameSuffix1;
 
+local nameSuffix = if (params.nameSuffix=="") then nameSuffix2 else params.nameSuffix;
+
 // Create a list of dictionary.c
 // Each item is a dictionary describing one step in the graph.
 local dagTemplates = [
@@ -226,6 +234,7 @@ local dagTemplates = [
         "--use_basic_auth=" + params.useBasicAuth,
         "--use_istio=" + params.useIstio,
         "--config_path=" + params.configPath,
+        "--cluster_creation_script=" + cluster_creation_script,
         // Increase the log level so that info level log statements show up.
         "--log-cli-level=info",        
         "--junitxml=" + artifactsDir + "/junit_kfctl-build-test" + nameSuffix + ".xml",
@@ -312,6 +321,7 @@ local deleteStep = if deleteKubeflow then
         "--junitxml=" + artifactsDir + "/junit_kfctl-go-delete-test.xml",
         "--app_path=" + appDir,
         "--kfctl_path=" + kfCtlPath,
+        "--cluster_deletion_script=" + cluster_deletion_script,
       ],
       working_dir=srcDir+ "/testing/kfctl",
     ),

diff --git a/testing/workflows/components/params.libsonnet b/testing/workflows/components/params.libsonnet
@@ -43,6 +43,9 @@
       useIstio: "true",
       testEndpoint: "false",
       configPath: "bootstrap/config/kfctl_gcp_iap_master.yaml",
+      cluster_creation_script: "",
+      cluster_deletion_script: "",
+      nameSuffix: "",
     },
     click_deploy_test: {
       bucket: "kubeflow-ci_temp",

diff --git a/testing/workflows/components/workflows.libsonnet b/testing/workflows/components/workflows.libsonnet
@@ -317,7 +317,7 @@
             "jupyter_test.py",
             // Test timeout in seconds.
             "--namespace=" + tests.stepsNamespace,
-            "--timeout=500",
+            "--timeout=1000",
             "--junitxml=" + tests.artifactsDir + "/junit_jupyter-test.xml",
           ],
           workingDir: tests.srcDir + "/kubeflow/jupyter/tests",