All Products
Search

This Product

    Container Service for Kubernetes:Configure the Arena client

    Document Center

    Container Service for Kubernetes:Configure the Arena client

    Last Updated:Mar 26, 2026

    Arena is a lightweight CLI for managing machine learning tasks on Kubernetes. It covers the full ML workflow—data preparation, model development, training, and prediction—improving the work efficiency of data scientists. Arena runs in deep learning frameworks optimized by Alibaba Cloud and integrates with Alibaba Cloud services to maximize the performance and utilization of heterogeneous computing resources through GPU sharing and Cloud Paralleled File System (CPFS) support.

    Prerequisites

    Before you begin, make sure you have:

    Step 1: Connect to the cluster

    Connection method depends on your cluster type.

    ACK managed clusters

    ACK managed clusters have no master nodes. Install the Arena client on your local machine (for example, a macOS computer) and connect via kubeconfig. See Connect to a cluster using kubectl.

    ACK dedicated clusters

    Use SSH to log in to a master node, then run Arena commands directly on that node. See Use SSH to connect to master nodes of an ACK dedicated cluster.

    Run kubectl get nodes to confirm your kubeconfig is configured correctly.

    Step 2: Install the Arena client

    Choose the package that matches your operating system and processor architecture, then run the commands to download, extract, and install.

    Linux/amd64

    # Download the Arena installation package.
wget https://aliacs-k8s-cn-hangzhou.oss-cn-hangzhou.aliyuncs.com/arena/arena-installer-0.12.0-linux-amd64.tar.gz

# Extract the package.
tar -zxvf arena-installer-0.12.0-linux-amd64.tar.gz

# Install Arena.
cd arena-installer-0.12.0-linux-amd64
bash install.sh --only-binary

    Linux/arm64

    # Download the Arena installation package.
wget https://aliacs-k8s-cn-hangzhou.oss-cn-hangzhou.aliyuncs.com/arena/arena-installer-0.12.0-linux-arm64.tar.gz

# Extract the package.
tar -zxvf arena-installer-0.12.0-linux-arm64.tar.gz

# Install Arena.
cd arena-installer-0.12.0-linux-arm64
bash install.sh --only-binary

    macOS/amd64

    # Download the Arena installation package.
wget https://aliacs-k8s-cn-hangzhou.oss-cn-hangzhou.aliyuncs.com/arena/arena-installer-0.12.0-darwin-amd64.tar.gz

# Extract the package.
tar -zxvf arena-installer-0.12.0-darwin-amd64.tar.gz

# Install Arena.
cd arena-installer-0.12.0-darwin-amd64
bash install.sh --only-binary

    macOS/arm64

    # Download the Arena installation package.
wget https://aliacs-k8s-cn-hangzhou.oss-cn-hangzhou.aliyuncs.com/arena/arena-installer-0.12.0-darwin-arm64.tar.gz

# Extract the package.
tar -zxvf arena-installer-0.12.0-darwin-arm64.tar.gz

# Install Arena.
cd arena-installer-0.12.0-darwin-arm64
bash install.sh --only-binary

    Step 3: (Optional) Enable shell auto-completion

    Shell auto-completion lets you press Tab to complete partially typed Arena commands.

    Install the completion package

    CentOS or Linux

    sudo yum install bash-completion -y

    Debian or Ubuntu

    sudo apt-get install bash-completion

    macOS

    brew install bash-completion@2

    Enable auto-completion in your shell profile

    bash (Linux)

    echo "source <(arena completion bash)" >> ~/.bashrc
chmod u+x ~/.bashrc

    bash (macOS)

    echo "source $(brew --prefix)/etc/profile.d/bash_completion.sh" >> ~/.bashrc

    Step 4: Verify the installation

    Run the following commands to confirm Arena is working.

    1. Check available GPU resources in the cluster:

      arena top node

      The output lists nodes and their GPU capacity. For example:

      NAME                        IPADDRESS      ROLE    STATUS  GPU(Total)  GPU(Allocated)
cn-huhehaote.192.168.X.XXX  192.168.0.117  <none>  ready   8           0
cn-huhehaote.192.168.X.XXX  192.168.0.118  <none>  ready   8           0
cn-huhehaote.192.168.X.XXX  192.168.0.119  <none>  ready   8           0
cn-huhehaote.192.169.X.XXX  192.168.0.120  <none>  ready   8           0
-----------------------------------------------------------------------------------------
Allocated/Total GPUs In Cluster:
0/32 (0%)

    2. Submit a test training job:

      arena submit tf \
      --name=firstjob \
      --gpus=1 \
      --image=registry.cn-hangzhou.aliyuncs.com/tensorflow-samples/tf-mnist-standalone:gpu \
      "python /app/main.py"

      Expected output:

      configmap/firstjob-tfjob created
configmap/firstjob-tfjob labeled
tfjob.kubeflow.org/firstjob created
INFO[0001] The Job firstjob has been submitted successfully
INFO[0001] You can run `arena get firstjob --type tfjob` to check the job status

    3. List all jobs:

      arena list

      Expected output:

      NAME      STATUS   TRAINER  AGE  NODE
firstjob  RUNNING  TFJOB    5s   192.168.X.XXX

    4. Check the job status:

      arena get firstjob

      Expected output:

      STATUS: SUCCEEDED
NAMESPACE: default
PRIORITY: N/A
TRAINING DURATION: 52s
NAME      STATUS     TRAINER  AGE  INSTANCE          NODE
firstjob  SUCCEEDED  TFJOB    14m  firstjob-chief-0  192.168.X.XXX

    5. View the job logs:

      arena logs --tail=10 firstjob

      Expected output:

      Accuracy at step 910: 0.9694
Accuracy at step 920: 0.9687
Accuracy at step 930: 0.9676
Accuracy at step 940: 0.9678
Accuracy at step 950: 0.9704
Accuracy at step 960: 0.9692
Accuracy at step 970: 0.9721
Accuracy at step 980: 0.9696
Accuracy at step 990: 0.9675
Adding run metadata for 999

      The job completed successfully.

    What's next