Installing Spark

Prerequisites

A Linux server that can access the public network is available. The recommended node specifications are 4 vCPUs and 8 GiB memory or higher.

Configuring the JDK

This section uses CentOS as an example to describe how to install JDK 1.8.

Obtain the available version.
```
yum -y list java*
```

Install JDK 1.8.

yum install -y java-1.8.0-openjdk java-1.8.0-openjdk-devel

Check the version after the installation.

# java -version
openjdk version "1.8.0_382"
OpenJDK Runtime Environment (build 1.8.0_382-b05)
OpenJDK 64-Bit Server VM (build 25.382-b05, mixed mode)

Add environment variables.
1. Linux environment variables are configured in the /etc/profile file.
```
vim /etc/profile
```
2. In the editing mode, add the following content to the end of the file:
```
JAVA_HOME=/usr/lib/jvm/java-1.8.0-openjdk-1.8.0.382.b05-1.el7_9.x86_64
PATH=$PATH:$JAVA_HOME/bin
CLASSPATH=.:$JAVA_HOME/lib/dt.jar:$JAVA_HOME/lib/tools.jar 
export JAVA_HOME  PATH  CLASSPATH
```
3. Save and close the profile file. Run the following command for the modification to take effect:
```
source /etc/profile
```
4. Check the JDK environment variables.
```
echo $JAVA_HOME
echo $PATH
echo $CLASSPATH
```

Obtaining the Spark Package

OBS matches Hadoop 2.8.3 and 3.1.1. Hadoop 3.1.1 is used in this example.

Download Spark v3.1.1. If Git is not installed, run yum install git to install it.
```
git clone -b v3.1.1 https://github.com/apache/spark.git
```
Modify the /dev/make-distribution.sh file and specify the Spark version so that the check can be skipped during compilation.
1. Search for the line where VERSION resides and check the number of the line where the version number is located.
```
cat ./spark/dev/make-distribution.sh |grep -n '^VERSION=' -A18
```
2. Comment out the content displayed from lines 129 to 147 and specify the version.
```
sed -i '129,147s/^/#/g' ./spark/dev/make-distribution.sh
sed -i '148a VERSION=3.1.3\nSCALA_VERSION=2.12\nSPARK_HADOOP_VERSION=3.1.1\nSPARK_HIVE=1' ./spark/dev/make-distribution.sh
```

Download the dependency.

wget https://archive.apache.org/dist//maven/maven-3/3.6.3/binaries/apache-maven-3.6.3-bin.tar.gz
tar -zxvf apache-maven-3.6.3-bin.tar.gz && mv apache-maven-3.6.3 ./spark/build

Run the following command to perform compilation:

./spark/dev/make-distribution.sh --name hadoop3.1 --tgz -Pkubernetes -Pyarn -Dhadoop.version=3.1.1

Wait for the compilation to complete. After the compilation, the software package is named spark-3.1.3-bin-hadoop3.1.tgz.

Configuring the Runtime Environment for Spark

To simplify the operation, use the root user to place the compiled package spark-3.1.3-bin-hadoop3.1.tgz in the /root directory on the operation node.

Move the software package to the /root directory.
```
mv ./spark/spark-3.1.3-bin-hadoop3.1.tgz /root
```

Run the following command to install Spark:

tar -zxvf spark-3.1.3-bin-hadoop3.1.tgz
mv spark-3.1.3-bin-hadoop3.1 spark-obs
cat >> ~/.bashrc <<EOF
PATH=/root/spark-obs/bin:\$PATH
PATH=/root/spark-obs/sbin:\$PATH
export SPARK_HOME=/root/spark-obs
EOF

source ~/.bashrc

Run the following command where binary spark-submit is used to check the Spark version:
```
spark-submit --version
```

Interconnecting Spark with OBS

Obtain Huawei Cloud OBS JAR. The hadoop-huaweicloud-3.1.1-hw-45.jar package is used, which can be obtained from https://github.com/huaweicloud/obsa-hdfs/tree/master/release.
```
wget https://github.com/huaweicloud/obsa-hdfs/releases/download/v45/hadoop-huaweicloud-3.1.1-hw-45.jar
```

Copy the package to the corresponding directory.

cp hadoop-huaweicloud-3.1.1-hw-45.jar /root/spark-obs/jars/

Modify Spark configuration items. To interconnect Spark with OBS, add ConfigMaps for Spark as follows:

Obtain the AK/SK. For details, see Access Keys.

Change the values of AK_OF_YOUR_ACCOUNT, SK_OF_YOUR_ACCOUNT, and OBS_ENDPOINT to the actual values.

AK_OF_YOUR_ACCOUNT: indicates the AK obtained in the previous step.
SK_OF_YOUR_ACCOUNT: indicates the SK obtained in the previous step.
OBS_ENDPOINT: indicates the OBS endpoint. It can be obtained in Regions and Endpoints.

cp ~/spark-obs/conf/spark-defaults.conf.template ~/spark-obs/conf/spark-defaults.conf

cat >> ~/spark-obs/conf/spark-defaults.conf <<EOF
spark.hadoop.fs.obs.readahead.inputstream.enabled=true
spark.hadoop.fs.obs.buffer.max.range=6291456
spark.hadoop.fs.obs.buffer.part.size=2097152
spark.hadoop.fs.obs.threads.read.core=500
spark.hadoop.fs.obs.threads.read.max=1000
spark.hadoop.fs.obs.write.buffer.size=8192
spark.hadoop.fs.obs.read.buffer.size=8192
spark.hadoop.fs.obs.connection.maximum=1000
spark.hadoop.fs.obs.access.key=AK_OF_YOUR_ACCOUNT
spark.hadoop.fs.obs.secret.key=SK_OF_YOUR_ACCOUNT
spark.hadoop.fs.obs.endpoint=OBS_ENDPOINT
spark.hadoop.fs.obs.buffer.dir=/root/hadoop-obs/obs-cache
spark.hadoop.fs.obs.impl=org.apache.hadoop.fs.obs.OBSFileSystem
spark.hadoop.fs.obs.connection.ssl.enabled=false
spark.hadoop.fs.obs.fast.upload=true
spark.hadoop.fs.obs.socket.send.buffer=65536
spark.hadoop.fs.obs.socket.recv.buffer=65536
spark.hadoop.fs.obs.max.total.tasks=20
spark.hadoop.fs.obs.threads.max=20
spark.kubernetes.container.image.pullSecrets=default-secret
EOF

Pushing an Image to SWR

To run Spark tasks in Kubernetes, build a Spark container image of the same version and upload it to SWR. A Dockerfile file has been generated during compilation. Use this file to create an image and push it to SWR.

Create an image.

cd ~/spark-obs
docker build -t spark:3.1.3-obs --build-arg spark_uid=0 -f kubernetes/dockerfiles/spark/Dockerfile .

Upload the image.
1. (Optional) Log in to the SWR console, choose Organizations in the navigation pane, and click Create Organization in the upper right corner of the page.
  Skip this step if you already have an organization.
2. Choose My Images in the navigation pane and click Upload Through Client. On the page displayed, click Generate a temporary login command and click to copy the command.
3. Run the login command copied in the previous step on the node. If the login is successful, the message "Login Succeeded" will display.
4. Log in to the node where the image is created and run the login command.
```
Docker tag [{ Image name }:{Version name }] swr. ap-southeast-1.myhuaweicloud.com/{ Organization name }/{Image name }:{version name }
docker push swr.ap-southeast-1.myhuaweicloud.com/{Organization name }/{Image name }:{Version name }
```
  Record the image access address for later use.
  
  For example, record the IP address as swr.ap-southeast-1.myhuaweicloud.com/dev-container/spark:3.1.3-obs.

Configuring Spark History Server

Modify the ~/spark-obs/conf/spark-defaults.conf file, enable Spark event logging, and configure the OBS bucket name and directory.
```
cat >> ~/spark-obs/conf/spark-defaults.conf <<EOF
spark.eventLog.enabled=true
spark.eventLog.dir=obs://{bucket-name}/{log-dir}/
EOF
```
- spark.eventLog.enabled: indicates that Spark event logging is enabled if it is set to true.
- spark.eventLog.dir: indicates the OBS bucket name and path. The bucket is named in the format of obs://{bucket-name}/{log-dir}/, for example, obs://spark-sh1/history-obs/. Ensure that the OBS bucket name and directory are correct.

Modify the ~/spark-obs/conf/spark-env.sh file. If the file does not exist, run the command to copy the template as a file:

cp ~/spark-obs/conf/spark-env.sh.template ~/spark-obs/conf/spark-env.sh

cat >> ~/spark-obs/conf/spark-env.sh <<EOF
SPARK_HISTORY_OPTS="-Dspark.history.fs.logDirectory=obs://{bucket-name}/{log-dir}/"
EOF

The OBS address must be the same as that in spark-default.conf in the previous step.

Start Spark History Server.

start-history-server.sh

Information similar to the following is displayed:

starting org.apache.spark.deploy.history.HistoryServer, logging to /root/spark-obs/logs/spark-root-org.apache.spark.deploy.history.HistoryServer-1-spark-sh1.out