azure-lng-bench-ss.sh

#!/bin/bash

# This script is used as internalCustomStartupScript to setup the VM of the Azure Scale Set.
# This setup is for Centos 7.2 VM only, and involves:
# - setting up Docker and Docker-Compose
# - setting up ProActive Node service with PNP
# - starting the ProActive Node service

set -x

###############################
# Bash function used by the VM to send debug messages to the dedicated Azure Queue
###############################
function debug {
  DEBUGCONTENT=`echo $1 | base64 -w 0`
  DEBUGMESSAGE="<QueueMessage><MessageText>$DEBUGCONTENT</MessageText></QueueMessage>"
  curl -X POST -d "$DEBUGMESSAGE" "https://$STORAGEACCOUNT.queue.core.windows.net/debug/messages?$SASKEY"
}

# Prerequisites: Install Docker and Docker Compose, add 'activeeon' user to docker group
#apt-get update
#apt-get install -y \
#     apt-transport-https \
#     ca-certificates \
#     curl \
#     gnupg2 \
#     software-properties-common
#curl -fsSL https://download.docker.com/linux/$(. /etc/os-release; echo "$ID")/gpg | apt-key add -
#add-apt-repository \
#   "deb [arch=amd64] https://download.docker.com/linux/$(. /etc/os-release; echo "$ID") \
#   $(lsb_release -cs) \
#   stable"
#apt-get update
#apt-get install -y docker-ce
#groupadd docker
#usermod -aG docker activeeon
#curl -L https://github.com/docker/compose/releases/download/1.17.0/docker-compose-`uname -s`-`uname -m` -o /usr/local/bin/docker-compose
#chmod +x /usr/local/bin/docker-compose

###############################
# Downloads and install ProActive Node as a Systemd service
###############################

# Install dependencies
yum install epel-release -y
if ! yum -y install git wget curl jq vim; then
  sleep 10
  if ! yum -y install git wget curl jq vim; then
    >&2 echo "Fatal error: Unable to run yum"
    halt
  fi
fi

# Retrieve NS config
JSONCONFIG=`base64 -d /var/lib/waagent/CustomData`
echo $JSONCONFIG
export RMURL=`echo $JSONCONFIG | jq '.rmurl' -r`
export CREDVALUE=`echo $JSONCONFIG | jq '.credentials' -r`
export NODESOURCENAME=`echo $JSONCONFIG | jq '.nodesourcename' -r`
export STORAGEACCOUNT=`echo $JSONCONFIG | jq '.storageaccount' -r`
export SASKEY=`echo $JSONCONFIG | jq '.saskey' -r`
#export USERCUSTOMSCRIPTURL=`echo $JSONCONFIG | jq '.usercustomscripturl' -r`
#export EXTERNALSTORAGEACCOUNT=`echo $JSONCONFIG | jq '.externalstorageaccount' -r`

# debug
debug "$HOSTNAME INFO: CustomData read properly (as I can write this message)"

# Overidded custom script if specified
#if [ ! -z "$USERCUSTOMSCRIPTURL" ]; then
#  curl -X GET "$USERCUSTOMSCRIPTURL" > user_custom_script.sh
#  ./user_custom_script.sh
#  if [ $? -ne 0 ]; then
#    debug "$HOSTNAME FATAL: User custom script exited with error: $?"
#    halt #sleep 9999 #exit -1
#  fi
#fi

# PNPS or PAMR can be infered from RM URL
#wget https://gist.githubusercontent.com/bamedro/c6c348af7741a4e09b4239077f5b2d67/raw/node.properties
#PROPERTIES="-Dproactive.net.nolocal=true -Dproactive.communication.protocol=pnps -Dproactive.pnps.heartbeat_factor=10 -Dlog4j.configuration=file:/opt/proactive/node.properties"
PROPERTIES="-Dproactive.net.nolocal=true -Dproactive.communication.protocol=pnp -Dproactive.pnps.heartbeat_factor=10"

# Retrieve node.jar
#mkdir -p /opt/proactive
#curl -X GET "https://$STORAGEACCOUNT.blob.core.windows.net/nodefiles/node.jar?$SASKEY" > /opt/proactive/node.jar
#chown -R tibcodev /opt/proactive

# Install Java 8
#cd /opt/proactive
#wget --no-clobber https://s3.amazonaws.com/ci-materials/Latest_jre/jre-8u131-linux-x64.tar.gz
#tar zxf jre-8u131-linux-x64.tar.gz
#ln -s /opt/proactive/jre1.8.0_131 /opt/proactive/java
##/opt/proactive/java/bin/java

# Read Azure Queue to get node name Deletion is performed on NS Side when nodes are properly registered
msg=$(curl "https://$STORAGEACCOUNT.queue.core.windows.net/nodeconfig/messages?visibilitytimeout=300&$SASKEY")
JSONMSG=`echo $msg | grep -oP '<MessageText>\K[^<]+' | base64 -d`
msgId=`echo $msg | grep -oP '<MessageId>\K[^<]+'`
popReceipt=`echo $msg | grep -oP '<PopReceipt>\K[^<]+'`
NODEBASENAME=`echo $JSONMSG | jq '.nodebasename' -r`
NODEINSTANCES=`echo $JSONMSG | jq '.nodeinstances' -r`
if [ -z "$NODEBASENAME" ]; then
  debug "$HOSTNAME FATAL: Unable to retrieve node configuration from 'nodeconfig' queue"
  halt #sleep 9999 #exit -1
fi

# Immediately delete the message from queue.
# If the script fails then the node will be considered lost,
# this must be the default behaviour until Azure scaleSet provide
# automatic redeployment in case of failure.
curl -X DELETE "https://${STORAGEACCOUNT}.queue.core.windows.net/nodeconfig/messages/${msgId}?popreceipt=${popReceipt}&${SASKEY}"

# debug
NOW=`date`
debug "$HOSTNAME INFO: Start filling the Table on $NOW"

# Retrieve instance metadata
INSTANCE_NAME=$(curl -H Metadata:true "http://169.254.169.254/metadata/instance/compute?api-version=2017-12-01" | jq .name | sed 's/"//g')
INSTANCE_ID=$(echo -n "$INSTANCE_NAME" | rev | cut -d'_' -f1 | rev)

# update Azure Table
ENTITY="{'PartitionKey':'$HOSTNAME','RowKey':'$NODEBASENAME', 'NodesCount':'$NODEINSTANCES', 'InstanceId': '$INSTANCE_ID'}"
curl -H "Content-Type: application/json" -d "$ENTITY" -X POST "https://$STORAGEACCOUNT.table.core.windows.net/nodesperhost?$SASKEY"
if [ $? -ne 0 ]; then
  debug "$HOSTNAME FATAL: Unable to register the host into 'nodesperhost' table"
  halt #sleep 9999 #exit -1
fi
NOW=`date`
debug "$HOSTNAME INFO: Terminated to fill the Table on $NOW"

# Generate proactive-node systemd service unit file
cat > /etc/systemd/system/proactive-node.service <<EOL
[Unit]
After=sshd.service

[Service]
ExecStart=/tibco/proactive/bin/proactive-node ${PROPERTIES} -v ${CREDVALUE} -w ${NODEINSTANCES} -r ${RMURL} -n ${NODEBASENAME} -s ${NODESOURCENAME}
User=tibcodev

[Install]
WantedBy=default.target
EOL
chmod 664 /etc/systemd/system/proactive-node.service

# Generate proactive-node-jar systemd service unit file
cat > /etc/systemd/system/proactive-node-jar.service <<EOL
[Unit]
After=sshd.service

[Service]
ExecStart=/usr/bin/java -jar /opt/proactive/node.jar ${PROPERTIES} -v ${CREDVALUE} -w ${NODEINSTANCES} -r ${RMURL} -n ${NODEBASENAME} -s lngmem
User=tibcodev

[Install]
WantedBy=default.target
EOL
chmod 664 /etc/systemd/system/proactive-node-jar.service

# Install ProActive Node Service
systemctl daemon-reload
systemctl enable proactive-node.service
#systemctl enable proactive-node-jar.service
sysctl fs.inotify.max_user_watches=524288 # Support for large number of nodes

# TODO: this must be called ONLY if the node.jar execution succeed
# Until here, if script fails, another host could reuse this nodeconfig message.
# Once the service is started, if something goes wrong, this node configuration will not be reusable from another host
#curl -X DELETE "https://${STORAGEACCOUNT}.queue.core.windows.net/nodeconfig/messages/${msgId}?popreceipt=${popReceipt}&${SASKEY}"
#if [ $? -ne 0 ]; then
#  debug "$HOSTNAME FATAL: Unable to delete nodeconfig from the queue."
#  halt #sleep 9999 #exit -1
#fi

# Debug message posted on debug queue
IP=`hostname -i`
debug "$HOSTNAME INFO: Service is ready to start: $IP , $NODEBASENAME , $NODEINSTANCES"
###############################

###############################
# Prepare script engines
###############################

# Python
#curl -O https://bootstrap.pypa.io/get-pip.py
#python2 get-pip.py
#python3 get-pip.py
#python2 /usr/local/bin/pip install py4j
#python2 /usr/local/bin/pip install numpy
#python3 /usr/local/bin/pip install py4j
#python3 /usr/local/bin/pip install numpy

# R
#apt-get install -y libssl-dev libcurl4-openssl-dev
#echo "deb http://cran.univ-paris1.fr/bin/linux/debian jessie-cran34/" >> /etc/apt/sources.list
#apt-key adv --keyserver keys.gnupg.net --recv-key 'E19F5F87128899B192B1A2C2AD5F960A256A04AF'
#apt-get install -y r-base r-cran-rjava
#Rscript --slave --no-save --no-restore-history -e "install.packages(c('gtools','stringr','devtools'), repos=c('http://www.freestatistics.org/cran/'))"
#update-alternatives --install /usr/bin/java java /opt/proactive/java/bin/java 1
#update-alternatives --set java /opt/proactive/java/bin/java

# Scilab
#apt-get install -y scilab

# OpenMPI
#apt-get install -y mpi-default-bin

# LnG specific configuration
sed -i -e '/lnh/d' /etc/fstab
if ! grep -q 'tibco' /etc/fstab; then
    echo '10.0.3.4:/tibco /tibco nfs auto,nofail,noatime,nolock,intr,tcp,actimeo=1800 0 0' >> /etc/fstab
fi
if ! grep -q 'algodata' /etc/fstab; then
    echo '10.0.3.4:/algodata /algodata nfs auto,nofail,noatime,nolock,intr,tcp,actimeo=1800 0 0' >> /etc/fstab
fi
mount -a
sed -i -e 's/PasswordAuthentication no/ PasswordAuthentication yes/g' /etc/ssh/sshd_config
systemctl reload sshd
sed -i -e '/ALGO_TOP/d' /etc/environment
echo 'ALGO_TOP=/algodata/algo_top/algos2dev1/AlgoOne5/algo/TOP' >> /etc/environment
echo 'PATH=/usr/bin:$PATH' >> /etc/environment
echo 'echo \"$@\"' > /bin/algoservices
chmod +x /bin/algoservices
sleep 10

# Setup Java
unlink /usr/bin/java
ln -s /tibco/proactive/jre/bin/java /usr/bin/java
PATH=$PATH:/usr/bin

###############################
# Let's start the ProActive node service
###############################
systemctl start proactive-node.service
#systemctl start proactive-node-jar.service

###############################
# Add additional jar required for script engines
###############################
# prepare one-jar lib dir
#sleep 120
#cd /tmp
#git clone https://github.com/gheon/proactive-node-additional-jar.git
#mv proactive-node-additional-jar/jri*.jar proactive-node-additional-jar/pa-*.jar proactive-node-additional-jar/rengine-*.jar /tmp/node/lib/
#mv proactive-node-additional-jar/scheduler-examples*.jar /tmp/node/lib/