Add Accumulo Docker image

prestodev/accumulo/Dockerfile

@@ -0,0 +1,95 @@
+FROM prestodev/centos6-oj8:unlabelled
+
+ARG ACCUMULO_VERSION=1.7.4
+ARG HADOOP_VERSION=2.6.5
+ARG ZOOKEEPER_VERSION=3.4.14
+ARG IMAGE_ROOT_INSTALL=/usr/local/lib
+
+ENV HADOOP_HOME=$IMAGE_ROOT_INSTALL/hadoop
+ENV ACCUMULO_HOME=$IMAGE_ROOT_INSTALL/accumulo
+ENV ZOOKEEPER_HOME=$IMAGE_ROOT_INSTALL/zookeeper
+
+ARG HADOOP_CONF=$HADOOP_HOME/conf
+ENV HADOOP_PREFIX=$HADOOP_HOME
+
+# TODO Apache Archive is rate limited -- these should probably go in S3
+ARG BASE_URL=https://archive.apache.org/dist
+ARG ACCUMULO_BINARY_PATH=$BASE_URL/accumulo/$ACCUMULO_VERSION/accumulo-$ACCUMULO_VERSION-bin.tar.gz
+ARG HADOOP_BINARY_PATH=$BASE_URL/hadoop/core/hadoop-$HADOOP_VERSION/hadoop-$HADOOP_VERSION.tar.gz
+ARG ZOOKEEPER_BINARY_PATH=$BASE_URL/zookeeper/zookeeper-$ZOOKEEPER_VERSION/zookeeper-$ZOOKEEPER_VERSION.tar.gz
+
+RUN yum update -y && \
+    yum install -y gettext && \
+    yum clean all -y
+
+RUN mkdir -p $IMAGE_ROOT_INSTALL
+
+# Download Hadoop
+RUN wget --quiet $HADOOP_BINARY_PATH && \
+    tar -xf hadoop-$HADOOP_VERSION.tar.gz -C $IMAGE_ROOT_INSTALL  && \
+    rm hadoop-$HADOOP_VERSION.tar.gz && \
+    ln -s $IMAGE_ROOT_INSTALL/hadoop-$HADOOP_VERSION $HADOOP_HOME
+
+# Download Accumulo
+RUN wget --quiet $ACCUMULO_BINARY_PATH && \
+    tar -xf accumulo-$ACCUMULO_VERSION-bin.tar.gz -C $IMAGE_ROOT_INSTALL && \
+    rm accumulo-$ACCUMULO_VERSION-bin.tar.gz && \
+    ln -s $IMAGE_ROOT_INSTALL/accumulo-$ACCUMULO_VERSION $ACCUMULO_HOME
+
+# Copy sample Accumulo configuration files and update the instance.volumes property
+ENV CONF_SAMPLE_DIR=$ACCUMULO_HOME/conf/examples/3GB/standalone
+RUN cp \
+    $CONF_SAMPLE_DIR/accumulo-env.sh \
+    $CONF_SAMPLE_DIR/accumulo-site.xml \
+    $CONF_SAMPLE_DIR/auditLog.xml \
+    $CONF_SAMPLE_DIR/generic_logger.properties \
+    $CONF_SAMPLE_DIR/generic_logger.xml \
+    $CONF_SAMPLE_DIR/log4j.properties \
+    $CONF_SAMPLE_DIR/monitor_logger.properties \
+    $CONF_SAMPLE_DIR/monitor_logger.xml \
+    $ACCUMULO_HOME/conf && \
+    sed -i s/"<value><\/value>"/"<value>file:\/\/\/tmp\/accumulo<\/value>"/ $ACCUMULO_HOME/conf/accumulo-site.xml
+
+# Download Presto Accumulo Iterators
+RUN wget --quiet https://repo1.maven.org/maven2/io/prestosql/presto-accumulo-iterators/344/presto-accumulo-iterators-344.jar && \
+    mkdir -p $ACCUMULO_HOME/lib/ext && \
+    mv presto-accumulo-iterators-344.jar $ACCUMULO_HOME/lib/ext
+
+# Download ZooKeeper
+RUN wget --quiet $ZOOKEEPER_BINARY_PATH && \
+    tar -xf zookeeper-$ZOOKEEPER_VERSION.tar.gz -C $IMAGE_ROOT_INSTALL && \
+    rm zookeeper-$ZOOKEEPER_VERSION.tar.gz && \
+    ln -s $IMAGE_ROOT_INSTALL/zookeeper-$ZOOKEEPER_VERSION $ZOOKEEPER_HOME && \
+    cp $ZOOKEEPER_HOME/conf/zoo_sample.cfg $ZOOKEEPER_HOME/conf/zoo.cfg
+
+RUN mkdir -p /var/log/zookeeper /var/log/accumulo
+
+COPY ./files /
+
+# Initialize Accumulo which requires ZooKeeper to be started
+RUN ${ZOOKEEPER_HOME}/bin/zkServer.sh start && \
+    ${ACCUMULO_HOME}/bin/accumulo init --user root --password secret --instance-name default
+
+# ZooKeeper
+EXPOSE 2181
+# Accumulo Monitor
+EXPOSE 50095
+# Accumulo Monitor Logs
+EXPOSE 4560
+# Accumulo TabletServer
+EXPOSE 9997
+# Accumulo TabletServer Replication Service
+EXPOSE 10002
+# Accumulo GC
+EXPOSE 50091
+# Accumulo Master
+EXPOSE 9999
+# Accumulo Replication Service
+EXPOSE 10001
+# Accumulo Tracer
+EXPOSE 12234
+
+ENTRYPOINT ADDRESS=`hostname -I` supervisord -c /etc/supervisord.conf
+
+HEALTHCHECK --start-period=5m \
+  CMD $ACCUMULO_HOME/bin/accumulo org.apache.accumulo.server.util.ListInstances | grep default

prestodev/accumulo/files/etc/supervisord.conf

@@ -0,0 +1,22 @@
+[supervisord]
+logfile = /var/log/supervisord.log
+logfile_maxbytes = 50MB
+logfile_backups=10
+loglevel = info
+pidfile = /var/run/supervisord.pid
+nodaemon = true
+directory = /tmp
+strip_ansi = false
+user=root
+
+[unix_http_server]
+file = /tmp/supervisor.sock
+
+[rpcinterface:supervisor]
+supervisor.rpcinterface_factory = supervisor.rpcinterface:make_main_rpcinterface
+
+[supervisorctl]
+serverurl = unix:///tmp/supervisor.sock
+
+[include]
+files = /etc/supervisord.d/*.conf

prestodev/accumulo/files/etc/supervisord.d/accumulo-gc.conf

@@ -0,0 +1,8 @@
+[program:accumulo-gc]
+command=/usr/local/lib/accumulo/bin/accumulo gc --address %(ENV_ADDRESS)s
+startsecs=2
+stopwaitsecs=10
+user=root
+redirect_stderr=true
+stdout_logfile=/var/log/accumulo/gc.log
+autostart=true

prestodev/accumulo/files/etc/supervisord.d/accumulo-master.conf

@@ -0,0 +1,8 @@
+[program:accumulo-master]
+command=/usr/local/lib/accumulo/bin/accumulo master --address %(ENV_ADDRESS)s
+startsecs=2
+stopwaitsecs=10
+user=root
+redirect_stderr=true
+stdout_logfile=/var/log/accumulo/master.log
+autostart=true

prestodev/accumulo/files/etc/supervisord.d/accumulo-monitor.conf

@@ -0,0 +1,8 @@
+[program:accumulo-monitor]
+command=/usr/local/lib/accumulo/bin/accumulo monitor --address %(ENV_ADDRESS)s
+startsecs=2
+stopwaitsecs=10
+user=root
+redirect_stderr=true
+stdout_logfile=/var/log/accumulo/monitor.log
+autostart=true

prestodev/accumulo/files/etc/supervisord.d/accumulo-tracer.conf

@@ -0,0 +1,8 @@
+[program:accumulo-tracer]
+command=/usr/local/lib/accumulo/bin/accumulo tracer --address %(ENV_ADDRESS)s
+startsecs=2
+stopwaitsecs=10
+user=root
+redirect_stderr=true
+stdout_logfile=/var/log/accumulo/tracer.log
+autostart=true

prestodev/accumulo/files/etc/supervisord.d/accumulo-tserver.conf

@@ -0,0 +1,8 @@
+[program:accumulo-tserver]
+command=/usr/local/lib/accumulo/bin/accumulo tserver --address %(ENV_ADDRESS)s
+startsecs=2
+stopwaitsecs=10
+user=root
+redirect_stderr=true
+stdout_logfile=/var/log/accumulo/tserver.log
+autostart=true

prestodev/accumulo/files/etc/supervisord.d/zookeeper.conf

@@ -0,0 +1,8 @@
+[program:zookeeper]
+command=/usr/local/lib/zookeeper/bin/zkServer.sh start-foreground
+startsecs=2
+stopwaitsecs=10
+user=root
+redirect_stderr=true
+stdout_logfile=/var/log/zookeeper/zookeeper.log
+autostart=true