*.*: Update README and image building utilities.

Remove pod-checkpoint-installer.

Author: Yifan Gu

cmd/checkpoint/README.md

@@ -79,3 +79,8 @@ ConfigMaps are stored using a path of:
 ```
 /etc/kubernets/checkpoint-configmaps/<namespace>/<pod-name>/<configmap-name>
 ```
+### Self Checkpointing
+
+The pod checkpoint will also checkpoint itself to the disk to handle the absence of the API server.
+After a node reboot, the on-disk pod-checkpointer will take over the responsibility.
+Once it reaches the API server and finds out that it's no longer being scheduled, it will clean up itself.

cmd/checkpoint/main.go

@@ -608,6 +608,13 @@ func handleRemove(remove []string) {
 		// Remove active checkpoints.
 		// We do this as the last step because we want to clean up
 		// resources before the checkpointer itself exits.
+		//
+		// TODO(yifan): Removing the pods after removing the secrets/configmaps
+		// might disturb other pods since they might want to use the configmap
+		// or secrets during termination.
+		//
+		// However, since we are not waiting for them to terminate anyway, so it's
+		// ok to just leave as is for now. We can handle this more gracefully later.
 		p = PodFullNameToActiveCheckpointPath(id)
 		if err := os.Remove(p); err != nil && !os.IsNotExist(err) {
 			glog.Errorf("Failed to remove active checkpoint %s: %v", p, err)

image/checkpoint/Dockerfile

@@ -1,5 +1,3 @@
 FROM alpine
 
 COPY checkpoint /checkpoint
-COPY checkpoint-install.sh /checkpoint-installer.sh
-COPY checkpoint-pod.yaml /checkpoint-pod.yaml

image/checkpoint/build-image.sh

@@ -12,9 +12,6 @@ function image::build() {
     # Add assets for container build
     cp ${BOOTKUBE_ROOT}/_output/bin/linux/checkpoint ${TEMP_DIR}
     cp ${BOOTKUBE_ROOT}/image/checkpoint/Dockerfile ${TEMP_DIR}
-    cp ${BOOTKUBE_ROOT}/image/checkpoint/checkpoint-install.sh ${TEMP_DIR}
-    cp ${BOOTKUBE_ROOT}/image/checkpoint/checkpoint-pod.yaml ${TEMP_DIR}
-    sed -i "s#{{ REPO }}:{{ TAG }}#${IMAGE_REPO}:${VERSION}#" ${TEMP_DIR}/checkpoint-pod.yaml
 
     docker build -t ${IMAGE_REPO}:${VERSION} -f ${TEMP_DIR}/Dockerfile ${TEMP_DIR}
     rm -rf ${TEMP_DIR}

image/checkpoint/checkpoint-install.sh

@@ -38,7 +38,7 @@ const (
 	AssetPathKubeDNSDeployment           = "manifests/kube-dns-deployment.yaml"
 	AssetPathKubeDNSSvc                  = "manifests/kube-dns-svc.yaml"
 	AssetPathSystemNamespace             = "manifests/kube-system-ns.yaml"
-	AssetPathCheckpointer                = "manifests/pod-checkpoint-installer.yaml"
+	AssetPathCheckpointer                = "manifests/pod-checkpointer.yaml"
 	AssetPathEtcdOperator                = "manifests/etcd-operator.yaml"
 	AssetPathEtcdSvc                     = "manifests/etcd-service.yaml"
 	AssetPathKenc                        = "manifests/kenc.yaml"

pkg/asset/asset.go

@@ -247,31 +247,61 @@ spec:
 	CheckpointerTemplate = []byte(`apiVersion: "extensions/v1beta1"
 kind: DaemonSet
 metadata:
-  name: checkpoint-installer
+  name: pod-checkpointer
   namespace: kube-system
   labels:
-    k8s-app: pod-checkpoint-installer
+    k8s-app: pod-checkpointer
 spec:
   template:
     metadata:
       labels:
-        k8s-app: pod-checkpoint-installer
+        k8s-app: pod-checkpointer
+      annotations:
+        checkpointer.alpha.coreos.com/checkpoint: "true"
     spec:
       nodeSelector:
         master: "true"
       hostNetwork: true
       containers:
-      - name: checkpoint-installer
-        image: quay.io/coreos/pod-checkpointer:417b8f7552ccf3db192ba1e5472e524848f0eb5f
+      - name: checkpoint
+        image: quay.io/coreos/pod-checkpointer:f0631b5e25a21db9c68cff6c5e719c72c0181c4f
         command:
-        - /checkpoint-installer.sh
+        - /checkpoint
+        - --v=4
+        - --lock-file=/var/run/lock/pod-checkpointer.lock
+        env:
+        - name: NODE_NAME
+          valueFrom:
+            fieldRef:
+              fieldPath: spec.nodeName
+        - name: POD_NAME
+          valueFrom:
+            fieldRef:
+              fieldPath: metadata.name
+        - name: POD_NAMESPACE
+          valueFrom:
+            fieldRef:
+              fieldPath: metadata.namespace
+        imagePullPolicy: Always
         volumeMounts:
-        - mountPath: /etc/kubernetes/manifests
-          name: etc-k8s-manifests
+        - mountPath: /etc/kubernetes
+          name: etc-kubernetes
+        - mountPath: /srv/kubernetes
+          name: srv-kubernetes
+        - mountPath: /var/run
+          name: var-run
+      hostNetwork: true
+      restartPolicy: Always
       volumes:
-      - name: etc-k8s-manifests
+      - name: etc-kubernetes
+        hostPath:
+          path: /etc/kubernetes
+      - name: srv-kubernetes
+        hostPath:
+          path: /srv/kubernetes
+      - name: var-run
         hostPath:
-          path: /etc/kubernetes/manifests
+          path: /var/run
 `)
 	ControllerManagerTemplate = []byte(`apiVersion: extensions/v1beta1
 kind: Deployment