Merge pull request #2 from mattmattox/manager-worker

adding manager and worker

Dockerfile → manager/Dockerfile

@@ -10,8 +10,7 @@ RUN apt-get update && apt-get install -yq --no-install-recommends \
     && apt-get clean && rm -rf /var/lib/apt/lists/*
 
 ## Install kubectl
-ADD kubectl /usr/local/bin/kubectl
-RUN chmod +x /usr/local/bin/kubectl
+RUN curl -LO "https://storage.googleapis.com/kubernetes-release/release/$(curl -s https://storage.googleapis.com/kubernetes-release/release/stable.txt)/bin/linux/amd64/kubectl" && chmod u+x kubectl && mv kubectl /usr/local/bin/kubectl
 
 ## Setup run script
 WORKDIR /root

run.sh → manager/run.sh

@@ -1,10 +1,10 @@
 #!/bin/bash
 
-if [[ -z $NODE_TIMEOUT ]]
+if [[ -z $nodeTimeout ]]
 then
-	NODE_TIMEOUT=360
+	nodeTimeout=360
 fi
-echo "Node timeout: $NODE_TIMEOUT"
+echo "nodeTimeout: $nodeTimeout"
 
 if [[ -z $AUTO_UNCORDON ]]
 then
@@ -35,7 +35,6 @@ do
 			echo "#########################################################"
 			echo "Checking $node"
 			current_status="$(kubectl get --no-headers $node | awk '{print $2}')"
-			##echo "Current node status: $current_status"
 			if [[ "$current_status" == "Ready" ]] || [[ "$current_status" == "Ready,SchedulingDisabled" ]]
 			then
 				echo "$node is ready"
@@ -73,7 +72,7 @@ do
 			                                mv ~/drained_nodes.tmp ~/drained_nodes
 							break
 						fi
-						if [ $count -gt $NODE_TIMEOUT ]
+						if [ $count -gt $nodeTimeout ]
 						then
 							echo "$node has been down for greater than 5Mins, assuming node is down for good."
 							echo "Starting drain of node..."
@@ -113,11 +112,11 @@ do
 									while ! kubectl get pods -n cattle-system | grep ^'cattle-cluster-agent-' | awk '{print $3}' | grep "Running"
 									do
 										sleep 1
-                                                                                cattlecount=$((cattlecount+1))
-                                                                                if [ $cattlecount -gt 60 ]
-                                                                                then
-                                                                                        break
-                                                                                fi
+                    cattlecount=$((cattlecount+1))
+                    if [ $cattlecount -gt 60 ]
+                    then
+                      break
+                    fi
 									done
 									echo "Scaling back down to 1..."
 									kubectl scale --replicas=1 deployment/cattle-cluster-agent -n cattle-system

worker/Dockerfile

@@ -0,0 +1,20 @@
+FROM ubuntu:18.04
+
+MAINTAINER [email protected]
+
+ENV DEBIAN_FRONTEND=noninteractive
+
+RUN apt-get update && apt-get install -yq --no-install-recommends \
+    apt-utils \
+    curl \
+    docker-ce-cli \
+    && apt-get clean && rm -rf /var/lib/apt/lists/*
+
+## Install kubectl
+RUN curl -LO "https://storage.googleapis.com/kubernetes-release/release/$(curl -s https://storage.googleapis.com/kubernetes-release/release/stable.txt)/bin/linux/amd64/kubectl" && chmod u+x kubectl && mv kubectl /usr/local/bin/kubectl
+
+## Setup run script
+WORKDIR /root
+ADD run.sh /root/run.sh
+
+CMD /root/run.sh

worker/run.sh

@@ -0,0 +1,69 @@
+#!/bin/bash
+
+function check_node {
+  current_status=`kubectl get node --no-headers "$1" |awk '{print $2}'`
+  if [[ "$current_status" == "Ready" ]] || [[ "$current_status" == "Ready,SchedulingDisabled" ]]
+  then
+    return 0
+  else
+    return 1
+  fi
+}
+
+if [[ -z $nodeTimeout ]]
+then
+  nodeTimeout=60
+fi
+echo "nodeTimeout: $nodeTimeout"
+
+echo "nodeName: $nodeName"
+if [[ -z "$nodeName" ]]
+then
+  echo "Missing nodeName"
+  exit 1
+fi
+
+echo "Verifing Docker CLI access..."
+if ! docker info
+then
+  echo "Problem accessing Docker CLI"
+  exit 2
+fi
+
+while true;
+do
+  echo "Checking node status..."
+  if check_node $nodeName
+  then
+    echo "Node is ready"
+  else
+    echo "Node is Not ready, rechecking..."
+    count=0
+    while true
+    do
+      if ! check_node $nodeName
+      then
+        echo "Sleeping for $count seconds"
+        sleep 1
+        count=$((count+1))
+      else
+        echo "Node is now ready"
+        break
+      fi
+      if [ $count -gt $nodeTimeout ]
+      then
+        echo "Node has been down for greater then $nodeTimeout seconds, assuming node is down"
+        echo "Attempting node recovery"
+        echo "Restarting kubelet"
+        docker restart kubelet
+        echo "Sleeping..."
+        sleep 15
+        if check_node $nodeName
+        then
+          echo "Node has recovered"
+          break
+        fi
+      fi
+    done
+  fi
+done