Installation Guide

Installation Overview

Network Planning

OmniStream adopts a single-node, containerized deployment model, running Flink within Docker containers.

A total of three Docker containers are deployed, whose specifications are all 8c32g. One container runs the Job Manager and the other two run the Task Manager. Figure 1 shows the networking diagram.

Figure 1 Networking diagram

Environment Requirements

Before installing OmniStream, prepare the hardware and software environments to facilitate subsequent installation operations.

Hardware Requirements

Table 1 Hardware requirements lists the hardware requirements.

Table 1 Hardware requirements

Item	Node
Processor	New Kunpeng 920 processor model
Memory size	384 GB (12 x 32 GB)
Memory frequency	2666MHz
Network	Service network: 10GE Management network: 1GE
Drive	System drive: 1 x RAID 0 (1 x 1.2 TB SAS HDD) Data drive: 12 x RAID 0 (12 x 8 TB SATA HDD)
RAID controller card	LSI SAS3508

OS and Software Requirements

Table 2 OS and software requirements lists the OS and software requirements.

Table 2 OS and software requirements

Item	Version	Description
OS	openEuler 22.03 LTS SP4	None
JDK	BiSheng JDK 1.8 (BiSheng JDK 1.8.0_342 recommended)	Deploy the JDK in all containers.
Flink	1.16.3 1.17.1 1.20.0	See Flink Deployment Guide (CentOS & openEuler). Deploy Flink in all containers.
Docker	19.03.15	None
Nexmark	v0.2.0	Perform compilation by following the official instructions. Deploy Nexmark in all containers.
Python	3.9.9	Deploy Python on the container or physical machine from which the job is submitted.
yaml-cpp	0.6.3	Deploy it on the container or physical machine from which the job is submitted.
GCC	10.3.1	None
Maven	3.8.7	Use it to generate the JAR package of UDF test cases.
Jemalloc	5.3.0	Use it to provide the header file used for UDF translation.
OmniOperator	master	Use it to provide the header file used for UDF translation.
Xxhash	0.8.2	Use it to provide the header file used for UDF translation.
nlohmann json	3.11.3	Use it to provide the header file used for UDF translation.

Obtaining the Software Packages

Table 3 OmniStream software packages describes the OmniStream software packages and how to obtain them.

Table 3 OmniStream software packages

Software Name	Package Name	Release Type	Description	How to Obtain
OmniStream package	BoostKit-omniruntime-omnistream-1.2.0.zip	Open source	OmniStream software installation package.	Link
UDF translator	UNT-1.0-35.noarch.rpm	Open source	UDF translator RPM package. After the installation is complete, the UDF translator is added to the /opt directory.	Link
AI4C	AI4C-1.0.4-8.aarch64.rpm	Open source	A framework that allows the compiler to integrate machine learning–driven optimization technologies. Install the RPM package.	Link
KACC_JSON	BoostKit-kaccjson_1.1.0.zip	Closed source	Self-developed C++ implementation package used to replace GSON in UDF translation. This ZIP package contains the adaptation layer and KACC_JSON implementation, and also contains header files and a static library. Obtain the Dependency_library_OmniStream.zip file and decompress it.	Link
KSL	BoostKit-ksl_2.5.1.zip	Closed source	Regular expression acceleration library, which contains the ReplaceAll function for optimizing the basic string library and contains header files and a static library.	Contact Huawei technical support.
Dependency Library	Dependency_library_Default	Open source	Library file on which OmniStream depends. Obtain the Dependency_library_OmniStream.zip file and decompress it.	Link 1

Verifying the Software Package Integrity

After downloading a software package from the Kunpeng community, verify the software package to ensure that it is consistent with the original one on the website.

Verify a software package as follows:

Obtain the digital certificate and software.
Obtain the verification tool and guide.
Verify the package integrity by following the procedure described in the OpenPGP Signature Verification Guide obtained from the URL.

Installing the Feature

Installing the Basic Environment

Installing Docker

Install Docker and deploy multiple containers to set up the Flink environment. If the server cannot connect to the Internet, configure a local yum repository according to your environment to ensure a smooth installation.

Install Docker and import the basic image. For details, see Docker Installation Guide (CentOS & openEuler).

cd /opt
wget --no-check-certificate https://mirrors.huaweicloud.com/openeuler/openEuler-22.03-LTS-SP4/docker_img/aarch64/openEuler-docker.aarch64.tar.xz
docker load < openEuler-docker.aarch64.tar.xz

Create a network in bridge mode.
```
docker network create -d bridge flink-network
```
Check whether the network is created successfully.
```
docker network ls
```

Create and start three Docker containers.

The containers have the 8c32g specification, and are named flink_jm_8c32g, flink_tm1_8c32g, and flink_tm2_8c32g. After all containers are started, the command execution process automatically exits.

docker run -it -d --name flink_jm_8c32g --cpus=8 --memory=32g --network flink-network openeuler-22.03-lts-sp4 /bin/bash 
docker run -it -d --name flink_tm1_8c32g --cpus=8 --memory=32g --network flink-network openeuler-22.03-lts-sp4 /bin/bash 
docker run -it -d --name flink_tm2_8c32g --cpus=8 --memory=32g --network flink-network openeuler-22.03-lts-sp4 /bin/bash

Query the container IDs.
```
docker ps
```
The first column in the following figure shows the container IDs.
Log in to all containers, enable the SSH service in the containers, and configure password-free login.
1. Log in to the containers in sequence and perform 5.b to 5.g.
```
docker exec -it flink_jm_8c32g /bin/bash
docker exec -it flink_tm1_8c32g /bin/bash
docker exec -it flink_tm2_8c32g /bin/bash
```
2. Install the SSH service dependency, password change dependency, command editing dependency, command search dependency, and network management dependency. If other dependencies are missing in the image, install them by yourself.
```
yum -y install openssh-clients openssh-server passwd vim findutils net-tools libXext libXrender gcc cmake make gcc-c++ unzip wget libXtst
```
3. Generate an RSA key.
```
ssh-keygen -A
```
4. Enable the SSH service in the container.
```
/usr/sbin/sshd -D &
```
5. Set a password for the container.
```
# Set the password of the current user. Remember the password.
passwd
```
6. Generate an RSA key again. When a message is displayed, press Enter.
```
ssh-keygen -t rsa
```
7. Exit the container.
```
exit
```
8. On the flink_jm_8c32g container, enable SSH password-free login to other containers.
```
docker exec -it flink_jm_8c32g /bin/bash
ssh-copy-id -i ~/.ssh/id_rsa.pub root@flink_tm1_8c32g
ssh-copy-id -i ~/.ssh/id_rsa.pub root@flink_tm2_8c32g
```

Installing the JDK

Install and configure the BiSheng JDK to provide a runtime environment for the Flink cluster.

Go to the /usr/local path on the physical machine and download the bisheng-jdk-8u342-linux-aarch64.tar.gz file.

cd /usr/local
wget --no-check-certificate https://mirror.iscas.ac.cn/kunpeng/archive/compiler/bisheng_jdk/bisheng-jdk-8u342-linux-aarch64.tar.gz

Extract bisheng-jdk-8u342-linux-aarch64.tar.gz in the /usr/local path and change the owner and owner group of the extracted JDK directory to root.

tar -zxvf bisheng-jdk-8u342-linux-aarch64.tar.gz
chown -R root /usr/local/bisheng-jdk1.8.0_342
chgrp -R root /usr/local/bisheng-jdk1.8.0_342

Installing Flink

Deploy and configure Flink on physical machines to run in multiple Docker containers, allowing Flink jobs to be submitted and executed.

Go to the /usr/local directory on the physical machine and download the Flink software package.
```
cd /usr/local
wget --no-check-certificate https://archive.apache.org/dist/flink/flink-1.16.3/flink-1.16.3-bin-scala_2.12.tgz
```
NOTE: flink-1.16.3-bin-scala_2.12.tgz is an example of the Flink software package name. If you are using a different Flink version, replace the example with the actual package name.
Extract the flink-1.16.3-bin-scala_2.12.tgz package in the /usr/local path and create a soft link for later version replacement.
```
tar -zxvf flink-1.16.3-bin-scala_2.12.tgz
chown -R root:root flink-1.16.3
ln -s flink-1.16.3 flink
```
NOTE: flink-1.16.3 is an example of the Flink software directory. If you are using a different Flink version, replace the example with the actual directory.

Configure the masters and workers files of Flink.

Open the /usr/local/flink/conf/masters file.
```
vi /usr/local/flink/conf/masters
```
Press i to enter the insert mode and change the content of masters to the flink_jm_8c32g container ID 8081. For example:
```
4a376b30106b:8081
```
Press Esc, type :wq!, and press Enter to save the file and exit.
Open the /usr/local/flink/conf/workers file.
```
vi /usr/local/flink/conf/workers
```
Press i to enter the insert mode, and change the content of workers to the IDs of the flink_tm1_8c32g and flink_tm2_8c32g containers. For example, configure four TaskManagers for each of two Docker containers.
```
c3ddf10d0353
c3ddf10d0353
c3ddf10d0353
c3ddf10d0353
c3bbdbcc1ae1
c3bbdbcc1ae1
c3bbdbcc1ae1
c3bbdbcc1ae1
```
Press Esc, type :wq!, and press Enter to save the file and exit.
Open the /usr/local/flink/conf/flink-conf.yaml file.
```
vi /usr/local/flink/conf/flink-conf.yaml
```

Press i to enter the insert mode, replace the configuration with the following, and change the value of jobmanager.rpc.address to the ID of the flink_jm_8c32g container ID. It is recommended that the total number of slots be greater than the degree of parallelism. For example:

taskmanager.memory.process.size: 8G
jobmanager.rpc.address: 4a376b30106b
jobmanager.rpc.port: 6123
jobmanager.memory.process.size: 8G
taskmanager.memory.process.size: 8G
taskmanager.numberOfTaskSlots: 2
parallelism.default: 16
rest.port: 8081
rest.bind-port: 8081
io.tmp.dirs: /tmp
#pipeline.operator-chaining: true
#execution:runtime-mode: STREAMING
#jobmanager.execution.failover-strategy: region
#table.optimizer.agg-phase-strategy: ONE_PHASE
#pipeline.object-reuse: true
#==============================================================================
# JVM
#==============================================================================
# JVM options for GC
#env.java.opts: -verbose:gc -XX:NewRatio=3 -XX:+PrintGCDetails -XX:+PrintGCDateStamps -XX:ParallelGCThreads=4

Press Esc, type :wq!, and press Enter to save the file and exit.

Installing Nexmark

Install and configure Nexmark to verify and test Flink.

Download nexmark-flink.tgz to the /opt directory on the physical machine and extract it.

cd /opt
wget --no-check-certificate https://github.com/nexmark/nexmark/releases/download/v0.2.0/nexmark-flink.tgz
tar xzf nexmark-flink.tgz
mv nexmark-flink nexmark
chown -R root:root nexmark

Modify the Nexmark configuration file /opt/nexmark/conf/nexmark.yaml.

Open the file.
```
vi /opt/nexmark/conf/nexmark.yaml
```

Press i to enter the insert mode and modify the file as follows: Change the value of nexmark.metric.reporter.host to the ID of the flink_jm_8c32g container.

################################################################################
#  Licensed to the Apache Software Foundation (ASF) under one
#  or more contributor license agreements.  See the NOTICE file
#  distributed with this work for additional information
#  regarding copyright ownership.  The ASF licenses this file
#  to you under the Apache License, Version 2.0 (the
#  "License"); you may not use this file except in compliance
#  with the License.  You may obtain a copy of the License at
#
#      http://www.apache.org/licenses/LICENSE-2.0
#
#  Unless required by applicable law or agreed to in writing, software
#  distributed under the License is distributed on an "AS IS" BASIS,
#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
#  See the License for the specific language governing permissions and
# limitations under the License.
################################################################################

#==============================================================================
# Rest & web frontend
#==============================================================================

# The metric reporter server host.
nexmark.metric.reporter.host: 4a376b30106b
# The metric reporter server port.
nexmark.metric.reporter.port: 9098

#==============================================================================
# Benchmark workload configuration (events.num)
#==============================================================================

nexmark.workload.suite.100m.events.num: 50000000
nexmark.workload.suite.100m.tps: 10000000
nexmark.workload.suite.100m.queries: "q0,q1,q2,q3,q4,q5,q7,q8,q9,q10,q11,q12,q13,q14,q15,q16,q17,q18,q19,q20,q21,q22"
nexmark.workload.suite.100m.queries.cep: "q0,q1,q2,q3"
nexmark.workload.suite.100m.warmup.duration: 120s
nexmark.workload.suite.100m.warmup.events.num: 50000000
nexmark.workload.suite.100m.warmup.tps: 10000000

#==============================================================================
# Benchmark workload configuration (tps, legacy mode)
# Without events.num and with monitor.duration
# NOTE: The numerical value of TPS is unstable
#==============================================================================

# When to monitor the metrics, default 3min after job is started
# nexmark.metric.monitor.delay: 3min
# How long to monitor the metrics, default 3min, i.e. monitor from 3min to 6min after job is started
# nexmark.metric.monitor.duration: 3min

# nexmark.workload.suite.10m.tps: 10000000
# nexmark.workload.suite.10m.queries: "q0,q1,q2,q3,q4,q5,q7,q8,q9,q10,q11,q12,q13,q14,q15,q16,q17,q18,q19,q20,q21,q22"

#==============================================================================
# Workload for data generation
#==============================================================================

nexmark.workload.suite.datagen.tps: 1000000
nexmark.workload.suite.datagen.queries: "insert_kafka"
nexmark.workload.suite.datagen.queries.cep: "insert_kafka"

#==============================================================================
# Flink REST
#==============================================================================

flink.rest.address: 4a376b30106b
flink.rest.port: 8081

#==============================================================================
# Kafka config
#==============================================================================

# kafka.bootstrap.servers: ***:9092

Press Esc, type :wq!, and press Enter to save the file and exit.

Installing Python

Install the Python RPM package in the flink_jm_8c32g container.

wget --no-check-certificate https://repo.openeuler.org/openEuler-preview/openEuler-22.03-LTS-SP4-HP-preview/OS/aarch64/Packages/python3-setuptools-59.4.0-5.oe2203sp4.noarch.rpm
rpm -ivh python3-setuptools-59.4.0-5.oe2203sp4.noarch.rpm
rm -rf /usr/bin/python
ln -s /usr/bin/python3 /usr/bin/python

Installing yaml-cpp

Install the yaml-cpp RPM package in the flink_jm_8c32g container.

wget --no-check-certificate https://repo.openeuler.org/openEuler-preview/openEuler-22.03-LTS-SP4-HP-preview/OS/aarch64/Packages/yaml-cpp-0.6.3-2.oe2203sp4.aarch64.rpm
rpm -ivh yaml-cpp-0.6.3-2.oe2203sp4.aarch64.rpm

Installing Necessary Dependencies

Install other software packages on which the feature depends.

Install the JSON dependency package on the physical machine.

wget -P /usr/local/flink/lib/ https://repo.maven.apache.org/maven2/org/json/json/20240303/json-20240303.jar

Install the Gson dependency package on the physical machine.

wget -P /usr/local/flink/lib/ https://repo.maven.apache.org/maven2/com/google/code/gson/gson/2.11.0/gson-2.11.0.jar

Install other dependency packages.
1. Obtain the dependency packages Dependency_library_OmniStream.zip obtained from Table 3 OmniStream software packages and extract it to the /opt directory.
```
unzip Dependency_library_OmniStream.zip -d /opt
cp /opt/Dependency_library_Default /opt/Dependency_library  
chmod -R 550 /opt/Dependency_library/*
```
2. Obtain nexmark-flink-0.3-SNAPSHOT.jar from Dependency_library, copy it to the flink/lib directory, and use it to replace nexmark-flink-0.2-SNAPSHOT.jar in nexmark/lib.
```
cp /opt/Dependency_library/nexmark-flink-0.3-SNAPSHOT.jar /usr/local/flink/lib
rm -rf /opt/nexmark/lib/nexmark-flink-0.2-SNAPSHOT.jar
cp /opt/Dependency_library/nexmark-flink-0.3-SNAPSHOT.jar /opt/nexmark/lib/
```
  After the preceding operations are complete, the flink/lib directory is scanned and the dependency package is loaded automatically during Flink startup. No additional installation commands are required.

Installing OmniStream

In independent deployment mode, you can install the precompiled OmniStream binary package and integrate it into Flink as a plugin.

Create the /usr/local/OmniStream directory on the physical machine to store OmniStream binary files.

Extract the BoostKit-omniruntime-omnistream-1.2.0.zip installation package obtained from Table 3 OmniStream software packages to the /usr/local/OmniStream directory.
```
unzip BoostKit-omniruntime-omnistream-1.2.0.zip
mkdir -p /usr/local/OmniStream
cp -r OmniStream_Default/* /usr/local/OmniStream/
chmod -R 550 /usr/local/OmniStream/*
```

Check the extracted files.

ls

After the binary package is extracted, the JAR package, SO files, basic library, and include directory are obtained.

flink-tnel-0.1-SNAPSHOT.jar
libtnel.so
libboundscheck.so
libboostkit-omniop-codegen-2.1.0-aarch64.so
libboostkit-omniop-operator-2.1.0-aarch64.so
libboostkit-omniop-vector-2.1.0-aarch64.so
libre2.so.11
libbasictypes
include

Edit the Flink configuration file /usr/local/flink/bin/config.sh.
1. Open the file.
```
vi /usr/local/flink/bin/config.sh
```
2. Press i to enter the insert mode, find the constructFlinkClassPath() function, comment out the echo line, and add a new PATCH path.
```
# echo "$FLINK_CLASSPATH""$FLINK_DIST"
PATCH=/usr/local/OmniStream/flink-tnel-0.1-SNAPSHOT.jar
echo $PATCH:"$FLINK_CLASSPATH""$FLINK_DIST"
```
  The updated configuration file is as follows:
3. Press Esc, type :wq!, and press Enter to save the file and exit.
Edit the Flink configuration file /usr/local/flink/conf/flink-conf.yaml.
1. Open the file.
```
vi /usr/local/flink/conf/flink-conf.yaml
```
2. Press i to enter the insert mode. In env.java.opts, add the libtnel.so file path, that is, the path to the .so files extracted in 2.
```
env.java.opts: -Djava.library.path=/usr/local/OmniStream/
```
3. Press Esc, type :wq!, and press Enter to save the file and exit.

Installing the UDF Translator

Installing the UDF Translator RPM Package

Obtain the UDF translator RPM package from Table 3 OmniStream software packages and upload it to the /opt directory on the physical machine.
Install the RPM package in the flink_jm_8c32g container. The default installation path is /opt.
```
rpm -ivh UNT-1.0-35.noarch.rpm
```
After the command is executed, the /opt/udf-trans-opt directory is automatically created.
NOTE: If the /opt/udf-trans-opt directory is not created, run the following command to manually create it:
```
mkdir /opt/udf-trans-opt
```
Copy the basic library directory installed in Installing OmniStream to the /opt/udf-trans-opt directory.
```
docker cp /usr/local/OmniStream/libbasictypes flink_jm_8c32g:/opt/udf-trans-opt
```

Installing UDF Dependencies

Upload and extract KACC_JSONUpload and decompress the KACC_JSON package on the physical machine..

Obtain the BoostKit-kacccjson_1.1.0.zip package from Table 3 OmniStream software packages, upload it to the /opt directory on the physical machine, and extract it.
```
cd /opt/
unzip BoostKit-kacccjson_1.1.0.zip
```

Copy the KACC_JSON header files and static libraries to the /opt/udf-trans-opt/libbasictypes directory of the UDF translator in the flink_jm_8c32g container.

docker exec flink_jm_8c32g mkdir -p /opt/udf-trans-opt/libbasictypes/include 
docker exec flink_jm_8c32g mkdir -p /opt/udf-trans-opt/libbasictypes/lib
docker cp include/kacc_json flink_jm_8c32g:/opt/udf-trans-opt/libbasictypes/include/
docker cp include/kacc_gson_shell flink_jm_8c32g:/opt/udf-trans-opt/libbasictypes/include/
docker cp libkaccgson.a flink_jm_8c32g:/opt/udf-trans-opt/libbasictypes/lib/

Download and extract the KSL installation package.
1. Obtain the BoostKit-ksl_2.5.1.zip package from Table 3 OmniStream software packages, upload it to the /opt directory on the physical machine, extract it, and copy it to the flink_jm_8c32g container.
```
cd /opt
unzip BoostKit-ksl_2.5.1.zip
docker cp boostkit-ksl-2.5.1-1.aarch64.rpm flink_jm_8c32g:/opt/udf-trans-opt/
```
2. Enter the flink_jm_8c32g container and install the KSL RPM package.
```
docker exec -it flink_jm_8c32g bash
cd /opt/udf-trans-opt/
rpm -ivh boostkit-ksl-2.5.1-1.aarch64.rpm
```

Install the OmniStream dependency header files on the physical machine.

Extract the depend.zip package obtained from Table 3 OmniStream software packages to the /usr/local/OmniStream/depend directory.

mkdir -p /usr/local/OmniStream/depend/
unzip depend.zip -d /usr/local/OmniStream/depend/
docker cp /usr/local/OmniStream/libbasictypes/OmniStream flink_jm_8c32g:/opt/udf-trans-opt/libbasictypes/include/
docker cp /usr/local/OmniStream/libbasictypes/include/core flink_jm_8c32g:/opt/udf-trans-opt/libbasictypes/include/
docker cp /usr/local/OmniStream/libbasictypes/include/runtime flink_jm_8c32g:/opt/udf-trans-opt/libbasictypes/include/
docker cp /usr/local/OmniStream/libbasictypes/include/streaming flink_jm_8c32g:/opt/udf-trans-opt/libbasictypes/include/
docker cp /usr/local/OmniStream/libbasictypes/include/table flink_jm_8c32g:/opt/udf-trans-opt/libbasictypes/include/
docker cp /usr/local/OmniStream/libbasictypes/include/third_party flink_jm_8c32g:/opt/udf-trans-opt/libbasictypes/include/

Install jemalloc.
1. Download jemalloc-5.3.0.tar.gz and upload it to the management node.
```
cd /opt/omni-operator/
tar zxvf jemalloc-5.3.0.tar.gz
mv jemalloc-5.3.0 jemalloc
```
  NOTE: The /opt/omni-operator/jemalloc directory can be customized.
2. Go to the jemalloc directory, run the script, and install the generated file.
```
cd jemalloc
./autogen.sh --disable-initial-exec-tls
make -j2
```
3. Copy /opt/omni-operator/jemalloc/lib/libjemalloc.so.2 to the /opt/omni-operator/lib directory.
```
cp /opt/omni-operator/jemalloc/lib/libjemalloc.so.2 /opt/omni-operator/lib/
```

Download OmniOperator and copy all the OmniOperator source code to the header file reference directory of the container UDF tool.

cd /usr/local/OmniStream/depend/
git clone https://atomgit.com/openeuler/OmniOperator.git -b master
mv OmniOperator OmniOperatorJIT 
docker cp /usr/local/OmniStream/depend/OmniOperatorJIT flink_jm_8c32g:/opt/udf-trans-opt/libbasictypes/include/

Download xxhash and copy xxhash.h to the header file reference directory of the UDF tool.

cd /usr/local/OmniStream/depend/
git clone https://github.com/Cyan4973/xxHash.git
cd xxHash && git checkout tags/v0.8.2
docker cp /usr/local/OmniStream/depend/xxhash.h flink_jm_8c32g:/opt/udf-trans-opt/libbasictypes/include/

Download nlohmann json and copy the include/nlohmann directory in the local code to the header file reference directory of the container UDF tool.

cd /usr/local/OmniStream/depend/
git clone https://github.com/nlohmann/json.git -b v3.11.3
docker cp /usr/local/OmniStream/depend/nlohmann/include/nlohmann flink_jm_8c32g:/opt/udf-trans-opt/libbasictypes/include/

Install the libboundscheck header file /opt/udf-trans-opt/libbasictypes.
1. Go to the flink_jm_8c32g container, create the /opt/udf-trans-opt/libbasictypes/include/libboundscheck directory, and exit the container.
```
docker exec -it flink_jm_8c32g bash
mkdir -p /opt/udf-trans-opt/libbasictypes/include/libboundscheck/
exit
```
2. Copy the include directory installed in Installing OmniStream to the /opt/udf-trans-opt/libbasictypes/include/libboundscheck directory in the flink_jm_8c32g container.
```
docker cp /usr/local/OmniStream/include flink_jm_8c32g:/opt/udf-trans-opt/libbasictypes/include/libboundscheck
```

Installing AI4C

Upload the dependency package AI4C-1.0.4-8.aarch64.rpm obtained from Table 3 OmniStream software packages to the /opt directory in the flink_jm_8c32g container.

docker cp AI4C-1.0.4-8.aarch64.rpm flink_jm_8c32g:/opt
docker exec -it flink_jm_8c32g bash
cd /opt
rpm -ivh --nodeps AI4C-1.0.4-8.aarch64.rpm

Deploying into Containers

After installing and configuring basic software on the physical machine, complete Installing the Basic Environment and Installing OmniStream to containerize the feature.

Copy the directories of Nexmark, JDK, Flink, third-party dependencies, and OmniStream to all containers.

docker cp /opt/nexmark flink_jm_8c32g:/usr/local/
docker cp /usr/local/flink-1.16.3 flink_jm_8c32g:/usr/local/
docker cp /usr/local/bisheng-jdk1.8.0_342 flink_jm_8c32g:/usr/local/
docker cp /opt/Dependency_library/Dependency_library_Default flink_jm_8c32g:/opt/Dependency_library
docker cp /usr/local/OmniStream flink_jm_8c32g:/usr/local/

docker cp /opt/nexmark flink_tm1_8c32g:/usr/local/
docker cp /usr/local/flink-1.16.3 flink_tm1_8c32g:/usr/local/
docker cp /usr/local/bisheng-jdk1.8.0_342 flink_tm1_8c32g:/usr/local/
docker cp /opt/Dependency_library/Dependency_library_Default flink_tm1_8c32g:/opt/Dependency_library
docker cp /usr/local/OmniStream flink_tm1_8c32g:/usr/local/

docker cp /opt/nexmark flink_tm2_8c32g:/usr/local/
docker cp /usr/local/flink-1.16.3 flink_tm2_8c32g:/usr/local/
docker cp /usr/local/bisheng-jdk1.8.0_342 flink_tm2_8c32g:/usr/local/
docker cp /opt/Dependency_library/Dependency_library_Default flink_tm2_8c32g:/opt/Dependency_library
docker cp /usr/local/OmniStream flink_tm2_8c32g:/usr/local/

Access the containers in sequence and perform 3.

docker exec -it flink_jm_8c32g /bin/bash
docker exec -it flink_tm1_8c32g /bin/bash
docker exec -it flink_tm2_8c32g /bin/bash

Set environment variables for Flink, JDK, Nexmark, and LLVM in each container.

Open the /etc/profile file.
```
vi /etc/profile
```

Press i to enter the insert mode and add the following content:

export JAVA_HOME=/usr/local/bisheng-jdk1.8.0_342
export PATH=$JAVA_HOME/bin:$PATH
export FLINK_HOME=/usr/local/flink-1.16.3
export PATH=$FLINK_HOME/bin:$PATH
export NEXMARK=/usr/local/nexmark
export PATH=$NEXMARK/bin:$PATH
export LD_PRELOAD=/opt/Dependency_library/libjemalloc.so.2:$LD_PRELOAD
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$JAVA_HOME/jre/lib/aarch64:$JAVA_HOME/lib:$JAVA_HOME/jre/lib/aarch64/server
export LD_LIBRARY_PATH=/usr/local/OmniStream:/opt/Dependency_library/:$LD_LIBRARY_PATH

Press Esc, type :wq!, and press Enter to save the file and exit.
Make the environment variables take effect.
```
source /etc/profile
```
Exit the container.
```
exit
```