feat: support adding init containers to vertices. Closes #284 (#285)

api/json-schema/schema.json

@@ -17271,6 +17271,13 @@
           "x-kubernetes-patch-merge-key": "name",
           "x-kubernetes-patch-strategy": "merge"
         },
+        "initContainers": {
+          "description": "List of init containers belonging to the pod. More info: https://kubernetes.io/docs/concepts/workloads/pods/init-containers/",
+          "items": {
+            "$ref": "#/definitions/io.k8s.api.core.v1.Container"
+          },
+          "type": "array"
+        },
         "limits": {
           "$ref": "#/definitions/io.numaproj.numaflow.v1alpha1.VertexLimits",
           "description": "Limits define the limitations such as buffer read batch size for all the vertices of a pipeline, will override pipeline level settings"
@@ -18701,6 +18708,13 @@
           "x-kubernetes-patch-merge-key": "name",
           "x-kubernetes-patch-strategy": "merge"
         },
+        "initContainers": {
+          "description": "List of init containers belonging to the pod. More info: https://kubernetes.io/docs/concepts/workloads/pods/init-containers/",
+          "items": {
+            "$ref": "#/definitions/io.k8s.api.core.v1.Container"
+          },
+          "type": "array"
+        },
         "interStepBufferServiceName": {
           "type": "string"
         },

api/openapi-spec/swagger.json

@@ -17279,6 +17279,13 @@
           "x-kubernetes-patch-merge-key": "name",
           "x-kubernetes-patch-strategy": "merge"
         },
+        "initContainers": {
+          "description": "List of init containers belonging to the pod. More info: https://kubernetes.io/docs/concepts/workloads/pods/init-containers/",
+          "type": "array",
+          "items": {
+            "$ref": "#/definitions/io.k8s.api.core.v1.Container"
+          }
+        },
         "limits": {
           "description": "Limits define the limitations such as buffer read batch size for all the vertices of a pipeline, will override pipeline level settings",
           "$ref": "#/definitions/io.numaproj.numaflow.v1alpha1.VertexLimits"
@@ -18683,6 +18690,13 @@
           "x-kubernetes-patch-merge-key": "name",
           "x-kubernetes-patch-strategy": "merge"
         },
+        "initContainers": {
+          "description": "List of init containers belonging to the pod. More info: https://kubernetes.io/docs/concepts/workloads/pods/init-containers/",
+          "type": "array",
+          "items": {
+            "$ref": "#/definitions/io.k8s.api.core.v1.Container"
+          }
+        },
         "interStepBufferServiceName": {
           "type": "string"
         },

docs/APIs.md

@@ -245,6 +245,20 @@ Settings for autoscaling
 </p>
 </td>
 </tr>
+<tr>
+<td>
+<code>initContainers</code></br> <em>
+<a href="https://v1-18.docs.kubernetes.io/docs/reference/generated/kubernetes-api/v1.18/#container-v1-core">
+\[\]Kubernetes core/v1.Container </a> </em>
+</td>
+<td>
+<em>(Optional)</em>
+<p>
+List of init containers belonging to the pod. More info:
+<a href="https://kubernetes.io/docs/concepts/workloads/pods/init-containers/">https://kubernetes.io/docs/concepts/workloads/pods/init-containers/</a>
+</p>
+</td>
+</tr>
 </tbody>
 </table>
 <h3 id="numaflow.numaproj.io/v1alpha1.Authorization">

docs/init-containers.md

@@ -0,0 +1,49 @@
+# Init Containers
+
+[Init Containers](https://kubernetes.io/docs/concepts/workloads/pods/init-containers/) can be provided for all the types of vertices.
+
+The following example shows how to add an init-container to a `udf` vertex.
+
+```yaml
+apiVersion: numaflow.numaproj.io/v1alpha1
+kind: Pipeline
+metadata:
+  name: my-pipeline
+spec:
+  vertices:
+    - name: my-udf
+      initContainers:
+        - name: my-init
+          image: busybox:latest
+          command: ["/bin/sh", "-c", "echo \"my-init is running!\" && sleep 60"]
+      udf:
+        container:
+          image: my-function:latest
+```
+
+The following example shows how to use init-containers and [`volumes`](./volumes.md) together to provide a `udf` container files on startup.
+```yaml
+apiVersion: numaflow.numaproj.io/v1alpha1
+kind: Pipeline
+metadata:
+  name: my-pipeline
+spec:
+  vertices:
+    - name: my-udf
+      volumes:
+        - name: my-udf-data
+          emptyDir: {}
+      initContainers:
+        - name: my-init
+          image: amazon/aws-cli:latest
+          command: ["/bin/sh", "-c", "aws s3 sync s3://path/to/my-s3-data /path/to/my-init-data"]
+          volumeMounts:
+            - mountPath: /path/to/my-init-data
+              name: my-udf-data
+      udf:
+        container:
+          image: my-function:latest
+          volumeMounts:
+            - mountPath: /path/to/my-data
+              name: my-udf-data
+```

docs/quick-start.md

@@ -41,7 +41,7 @@ simple-pipeline-cat-0-kqrhy                  2/2     Running     0          10s
 simple-pipeline-in-0-rhpjm                   1/1     Running     0          11s
 
 # Watch the log for the `output` vertex
-kubectl logs -f simple-pipeline-out-0-xxxx main
+kubectl logs -f simple-pipeline-out-0-xxxx
 2022/08/25 23:59:38 (out) {"Data":"VT+G+/W7Dhc=","Createdts":1661471977707552597}
 2022/08/25 23:59:38 (out) {"Data":"0TaH+/W7Dhc=","Createdts":1661471977707615953}
 2022/08/25 23:59:38 (out) {"Data":"EEGH+/W7Dhc=","Createdts":1661471977707618576}
@@ -96,12 +96,12 @@ curl -kq -X POST -d "103" https://localhost:8444/vertices/in
 curl -kq -X POST -d "104" https://localhost:8444/vertices/in 
 
 # Watch the log for the even vertex
-kubectl logs -f even-odd-even-sink-0-nf2ql main
+kubectl logs -f even-odd-even-sink-0-nf2ql
 2022/09/07 22:29:40 (even-sink) 102
 2022/09/07 22:29:40 (even-sink) 104
 
 # Watch the log for the odd vertex
-kubectl logs -f even-odd-odd-sink-0-a6p0n main
+kubectl logs -f even-odd-odd-sink-0-a6p0n
 2022/09/07 22:30:19 (odd-sink) 101
 2022/09/07 22:30:19 (odd-sink) 103
 ```

docs/user-defined-functions.md

@@ -59,3 +59,12 @@ Some environment variables are available in the user defined function Pods, they
 - `NUMAFLOW_REPLICA` - Replica index.
 - `NUMAFLOW_PIPELINE_NAME` - Name of the pipeline.
 - `NUMAFLOW_VERTEX_NAME` - Name of the vertex.
+
+### Configuration
+
+Configuration data can be provided to the UDF container at runtime multiple ways.
+* [`environment variables`](./environment-variables.md)
+* `args`
+* `command`
+* [`volumes`](./volumes.md)
+* [`init containers`](./init-containers.md)

mkdocs.yml

@@ -64,6 +64,7 @@ nav:
       - container-resources.md
       - volumes.md
       - environment-variables.md
+      - init-containers.md
       - watermarks.md
       - APIs.md
   - Operator Manual:

pkg/apis/numaflow/v1alpha1/const.go

@@ -46,7 +46,7 @@ const (
 
 	// container names.
 	CtrInit   = "init"
-	CtrMain   = "main"
+	CtrMain   = "numa"
 	CtrUdf    = "udf"
 	CtrUdsink = "udsink"
 

pkg/apis/numaflow/v1alpha1/container_builder_test.go

@@ -27,7 +27,7 @@ func Test_containerBuilder(t *testing.T) {
 			resources: testResources,
 		}).
 		build()
-	assert.Equal(t, "main", c.Name)
+	assert.Equal(t, "numa", c.Name)
 	assert.Len(t, c.VolumeMounts, 0)
 	assert.Equal(t, testResources, c.Resources)
 }