OnFailurePolicy docs & endtoend tests

.github/workflows/tests.yml

@@ -4,25 +4,24 @@ on:
     branches:
     - master
   pull_request:
-    branches:
-    - master
 jobs:
   end-to-end:
     runs-on: ubuntu-latest
-    strategy:
-      max-parallel: 4
-      matrix:
-        go-version: [1.10]
     steps:
     - uses: actions/checkout@v1
     - name: Kustomize and diff
       run: DELTA_CHECK=true make kustomize
-    - name: Set up Go@${{ matrix.go-version }}
-      uses: actions/setup-go@v1
-      with:
-        go-version: ${{ matrix.go-version }}
-    - name: Run end-to-end tests
-      run: make end2end
+    - uses: engineerd/[email protected]
+    - name: End2End
+      env:
+        DOCKER_USERNAME: ${{ github.actor }}
+        DOCKER_PASSWORD: "${{ secrets.GITHUB_TOKEN }}"
+      run: |
+        kubectl cluster-info
+        kubectl get pods -n kube-system
+        echo "current-context:" $(kubectl config current-context)
+        echo "environment-kubeconfig:" ${KUBECONFIG}
+        make end2end_execute
   docs:
     runs-on: ubuntu-latest
     strategy:

deployment/eks/flyte_generated.yaml

@@ -594,7 +594,7 @@ spec:
       labels:
         app: flyteadmin
         app.kubernetes.io/name: flyteadmin
-        app.kubernetes.io/version: 0.2.8
+        app.kubernetes.io/version: 0.2.9
     spec:
       containers:
       - command:
@@ -603,7 +603,7 @@ spec:
         - --config
         - /etc/flyte/config/flyteadmin_config.yaml
         - serve
-        image: docker.io/lyft/flyteadmin:v0.2.8
+        image: docker.io/lyft/flyteadmin:v0.2.9
         imagePullPolicy: IfNotPresent
         name: flyteadmin
         ports:
@@ -754,7 +754,7 @@ spec:
       labels:
         app: flytepropeller
         app.kubernetes.io/name: flytepropeller
-        app.kubernetes.io/version: 0.2.45
+        app.kubernetes.io/version: 0.2.63
     spec:
       containers:
       - args:
@@ -767,7 +767,7 @@ spec:
           valueFrom:
             fieldRef:
               fieldPath: metadata.name
-        image: docker.io/lyft/flytepropeller:v0.2.45
+        image: docker.io/lyft/flytepropeller:v0.2.63
         imagePullPolicy: IfNotPresent
         name: flytepropeller
         ports:

deployment/sandbox/flyte_generated.yaml

@@ -1024,7 +1024,7 @@ spec:
       labels:
         app: flyteadmin
         app.kubernetes.io/name: flyteadmin
-        app.kubernetes.io/version: 0.2.8
+        app.kubernetes.io/version: 0.2.9
     spec:
       containers:
       - command:
@@ -1033,7 +1033,7 @@ spec:
         - --config
         - /etc/flyte/config/flyteadmin_config.yaml
         - serve
-        image: docker.io/lyft/flyteadmin:v0.2.8
+        image: docker.io/lyft/flyteadmin:v0.2.9
         imagePullPolicy: IfNotPresent
         name: flyteadmin
         ports:
@@ -1191,7 +1191,7 @@ spec:
       labels:
         app: flytepropeller
         app.kubernetes.io/name: flytepropeller
-        app.kubernetes.io/version: 0.2.45
+        app.kubernetes.io/version: 0.2.63
     spec:
       containers:
       - args:
@@ -1206,7 +1206,7 @@ spec:
           valueFrom:
             fieldRef:
               fieldPath: metadata.name
-        image: docker.io/lyft/flytepropeller:v0.2.45
+        image: docker.io/lyft/flytepropeller:v0.2.63
         imagePullPolicy: IfNotPresent
         name: flytepropeller
         ports:

deployment/test/flyte_generated.yaml

@@ -644,7 +644,7 @@ spec:
       labels:
         app: flyteadmin
         app.kubernetes.io/name: flyteadmin
-        app.kubernetes.io/version: 0.2.8
+        app.kubernetes.io/version: 0.2.9
     spec:
       containers:
       - command:
@@ -653,7 +653,7 @@ spec:
         - --config
         - /etc/flyte/config/flyteadmin_config.yaml
         - serve
-        image: docker.io/lyft/flyteadmin:v0.2.8
+        image: docker.io/lyft/flyteadmin:v0.2.9
         imagePullPolicy: IfNotPresent
         name: flyteadmin
         ports:
@@ -772,7 +772,7 @@ spec:
       labels:
         app: flytepropeller
         app.kubernetes.io/name: flytepropeller
-        app.kubernetes.io/version: 0.2.45
+        app.kubernetes.io/version: 0.2.63
     spec:
       containers:
       - args:
@@ -785,7 +785,7 @@ spec:
           valueFrom:
             fieldRef:
               fieldPath: metadata.name
-        image: docker.io/lyft/flytepropeller:v0.2.45
+        image: docker.io/lyft/flytepropeller:v0.2.63
         imagePullPolicy: IfNotPresent
         name: flytepropeller
         ports:

end2end/tests/endtoend.yaml

@@ -11,7 +11,7 @@ spec:
     command:
     - bash
     - -c
-    image: docker.io/lyft/flytetester:v0.1.6
+    image: docker.io/lyft/flytetester:e29ac562f053741213efcead5950b4b8bc28cfcf
     imagePullPolicy: IfNotPresent
     name: flytetester
     resources:

kustomize/base/admindeployment/deployment.yaml

@@ -16,7 +16,7 @@ spec:
       labels:
         app: flyteadmin
         app.kubernetes.io/name: flyteadmin
-        app.kubernetes.io/version: 0.2.8
+        app.kubernetes.io/version: 0.2.9
       annotations:
         prometheus.io/scrape: "true"
         prometheus.io/port: "10254"
@@ -31,15 +31,15 @@ spec:
           name: flyte-admin-config
       initContainers:
       - name: run-migrations
-        image: docker.io/lyft/flyteadmin:v0.2.8
+        image: docker.io/lyft/flyteadmin:v0.2.9
         imagePullPolicy: IfNotPresent
         command: ["flyteadmin", "--logtostderr", "--config", "/etc/flyte/config/flyteadmin_config.yaml", "migrate", "run"]
         volumeMounts:
         - name: config-volume
           mountPath: /etc/flyte/config
       containers:
       - name: flyteadmin
-        image: docker.io/lyft/flyteadmin:v0.2.8
+        image: docker.io/lyft/flyteadmin:v0.2.9
         imagePullPolicy: IfNotPresent
         command: ["flyteadmin", "--logtostderr", "--config", "/etc/flyte/config/flyteadmin_config.yaml", "serve"]
         ports:

kustomize/base/propeller/deployment.yaml

@@ -15,7 +15,7 @@ spec:
       labels:
         app: flytepropeller
         app.kubernetes.io/name: flytepropeller
-        app.kubernetes.io/version: 0.2.45
+        app.kubernetes.io/version: 0.2.63
       annotations:
         prometheus.io/scrape: "true"
         prometheus.io/port: "10254"
@@ -31,7 +31,7 @@ spec:
           name: flyte-plugin-config
       containers:
       - name: flytepropeller
-        image: docker.io/lyft/flytepropeller:v0.2.45
+        image: docker.io/lyft/flytepropeller:v0.2.63
         command:
         - flytepropeller
         args:

rsts/user/features/index.rst

@@ -15,3 +15,4 @@ Flyte Features
     task_cache
     roles
     single_task_execution
+    on_failure_policy

rsts/user/features/on_failure_policy.rst

@@ -0,0 +1,45 @@
+.. _on-failuire-policy:
+
+What is it
+==========
+
+The default behavior for when a node fails in a workflow is to immediately abort the entire workflow. The reasoning behind this thinking
+is to avoid wasting resources since the workflow will end up failing anyway. There are certain cases however, when it's desired for the
+workflow to carry on executing the branches it can execute. 
+
+For example when the remaining tasks are marked as :ref:`cacheable <features-task_cache>`. 
+Once the failure has been fixed and the workflow is relaunched, cached tasks will be bypassed quickly.
+
+How to use it
+-------------
+
+Use on_failure attribute on workflow_class.
+
+.. code:: python
+
+   from flytekit.models.core.workflow import WorkflowMetadata
+
+   @workflow_class(on_failure=WorkflowMetadata.OnFailurePolicy.FAIL_AFTER_EXECUTABLE_NODES_COMPLETE)
+   class RunToCompletionWF(object):
+     pass
+
+Available values in the policy:
+
+.. code:: python
+
+    class OnFailurePolicy(object):
+        """
+        Defines the execution behavior of the workflow when a failure is detected.
+        Attributes:
+            FAIL_IMMEDIATELY                        Instructs the system to fail as soon as a node fails in the
+                                                    workflow. It'll automatically abort all currently running nodes and
+                                                    clean up resources before finally marking the workflow executions as failed.
+            FAIL_AFTER_EXECUTABLE_NODES_COMPLETE    Instructs the system to make as much progress as it can. The system
+                                                    will not alter the dependencies of the execution graph so any node 
+                                                    that depend on the failed node will not be run. Other nodes that will
+                                                    be executed to completion before cleaning up resources and marking
+                                                    the workflow execution as failed.
+        """
+
+        FAIL_IMMEDIATELY = _core_workflow.WorkflowMetadata.FAIL_IMMEDIATELY        
+        FAIL_AFTER_EXECUTABLE_NODES_COMPLETE = _core_workflow.WorkflowMetadata.FAIL_AFTER_EXECUTABLE_NODES_COMPLETE