Automate Harness Pod resolution into Spark

.circleci/config.yml

@@ -22,7 +22,8 @@ jobs:
           name: Setup dependencies
           command: |
             latests=$(wget -qO - https://raw.githubusercontent.com/${CIRCLE_PROJECT_USERNAME}/${CIRCLE_PROJECT_REPONAME}/master/.latests)
-            version=${CIRCLE_TAG#v}
+            version=2.3.3
+            # version=${CIRCLE_TAG#v}
             [ -n "$version" ] || version=$(echo "${latests}" | cut -f1 -d' ')
             printf 'export LATESTS=%s\n' ${latests} >> $BASH_ENV
             printf 'export VERSION=%s\n' ${version} >> $BASH_ENV
@@ -41,15 +42,14 @@ jobs:
           command: |
             SHORT_GIT_HASH=$(echo $CIRCLE_SHA1 | cut -c -7)
             DATE_BUILD=$(date +'%Y-%m-%d')
-
+            
             case "${CIRCLE_BRANCH}" in
               master|develop|devops)
-              docker image build -f Dockerfile \
+              docker build -t actionml/spark:latest --no-cache\
                 --build-arg release=${CIRCLE_TAG:-build} \
                 --build-arg version=${VERSION%%-*} \
                 --build-arg GIT_HASH=$SHORT_GIT_HASH --build-arg DATE_BUILD=$DATE_BUILD \
-                --build-arg BRANCH=${CIRCLE_BRANCH} \
-                -t actionml/spark:latest ./
+                --build-arg BRANCH=${CIRCLE_BRANCH} .
               ;;
               *)
               ;;

Dockerfile

@@ -8,7 +8,6 @@ ARG GIT_HASH
 ARG DATE_BUILD
 ARG BRANCH
 
-
 LABEL com.actionml.spark.vendor=ActionML \
       com.actionml.spark.version=$version \
       com.actionml.spark.release=$release
@@ -19,9 +18,11 @@ ENV DATE_BUILD=${DATE_BUILD}
 
 ENV SPARK_HOME=/spark \
     SPARK_PGP_KEYS="6EC5F1052DF08FF4 DCE4BFD807461E96"
+RUN apk update && apk add --no-cache curl jq
 
 RUN adduser -Ds /bin/bash -h ${SPARK_HOME} spark && \
     apk add --no-cache bash tini libc6-compat linux-pam krb5 krb5-libs && \
+
 # download dist
     apk add --virtual .deps --no-cache curl tar gnupg && \
     cd /tmp && export GNUPGHOME=/tmp && \
@@ -30,6 +31,7 @@ RUN adduser -Ds /bin/bash -h ${SPARK_HOME} spark && \
         https://archive.apache.org/dist/spark/spark-${version}/{${file},${file}.asc} && \
     gpg --keyserver hkp://pool.sks-keyservers.net --recv-keys ${SPARK_PGP_KEYS} && \
     gpg --batch --verify ${file}.asc ${file} && \
+
 # create spark directories
     mkdir -p ${SPARK_HOME}/work ${SPARK_HOME}/conf && chown spark:spark ${SPARK_HOME}/work && \
     tar -xzf ${file} --no-same-owner --strip-components 1 && \
@@ -38,10 +40,14 @@ RUN adduser -Ds /bin/bash -h ${SPARK_HOME} spark && \
     apk --no-cache del .deps && ls -A | xargs rm -rf
 
 COPY entrypoint.sh /
+COPY setHarnessHost.sh /harnesstool/
 COPY spark-env.sh ${SPARK_HOME}/conf/
 
+RUN chmod +x /harnesstool/setHarnessHost.sh && echo "* * * * * /bin/bash /harnesstool/setHarnessHost.sh" >> /etc/crontabs/root
+
+RUN curl https://repo1.maven.org/maven2/io/prometheus/jmx/jmx_prometheus_javaagent/0.12.0/jmx_prometheus_javaagent-0.12.0.jar --create-dirs -o ${SPARK_HOME}/prometheus/jmx_prometheus_javaagent-0.12.0.jar
+COPY config.yaml ${SPARK_HOME}/prometheus/conf/config.yaml 
+
 WORKDIR ${SPARK_HOME}/work
 ENTRYPOINT [ "/entrypoint.sh" ]
 
-# Specify the User that the actual main process will run as
-USER spark:spark

config.yaml

@@ -0,0 +1,7 @@
+---
+startDelaySeconds: 0
+ssl: false
+lowercaseOutputName: false
+lowercaseOutputLabelNames: false
+rules:
+  - pattern: ".*"

setHarnessHost.sh

@@ -0,0 +1,18 @@
+#!/bin/bash
+
+   POD_IP=`curl -s https://$KUBERNETES_SERVICE_HOST/api/v1/namespaces/default/pods/harness-0 --header "Authorization: Bearer $KUBE_TOKEN" --insecure | jq -r .status.podIP`
+
+   TEMP_HOSTS=/harnesstool/newhosts
+   echo "$(cat /etc/hosts)" > $TEMP_HOSTS
+   HOSTS=`cat $TEMP_HOSTS | grep harness-0`
+
+   if [ "$HOSTS" != "" ]
+   then
+      sed -i "s|.*.harness-0|$POD_IP \tharness-0|g" $TEMP_HOSTS
+   else
+      echo "$POD_IP harness-0" >> $TEMP_HOSTS
+   fi
+
+   echo "$(cat $TEMP_HOSTS)" > /etc/hosts
+
+   echo "$(date '+%Y-%m-%d %T') Haness-0 IP: $POD_IP" >> /harnesstool/harness-host.log