📦Container Management

AI-Stack provides a web-based interface for managing containers, giving users easy access to machine learning environments such as TensorFlow, PyTorch, and more.

Create Container

🔹 Single Container Creation

To create a single container:

1

Go to [Container List] and click the ➕ icon.

2

Fill in the following:

• Container Name

• Start Time

  • ⏱️ Execute Immediately or specify a future time

• End Time

  • N/A for no expiration

  • Or specify a time range using:

    • Select Time

    • Set Hours

    🧠 Use “Set Hours” to auto-delete the container after a specific duration.

3

Select:

• GPU Type and Quantity

• Hardware Specs (CPU cores, RAM)

• Container Image (public or custom)

4

(Optional) Set:

• Password or Key Pair (if SSH login is required)

• Shared Memory for multi-GPU parallel processing

• Mount Volume and default Jupyter workspace

5

Click [Next Step | Create] and review the configuration.

6

Confirm and click [Create].

✅ Once created, the container will appear under Container List and Project Detail, with status set to Running.

---

📚Batch Container Creation

Project managers can create multiple containers for team members:

1

Follow the Single Container Creation steps.

2

In the Advanced Settings, enable the [Batch] option.

3

Select members and move them from left ➡️ right block.

a. Specify the Quantity (containers per member).

4

Review and click [Create].

📌 Useful for workshops, training, or pre-provisioned resource

---

Container List View

Click a [Running] container to view:

• 📄 Deployment Detail

• 💾 Mounting Volume

• 🌐 External Volume

• 🧾 Description

• 📈 Monitor

• 🔔 Event Logs

---

Access Methods

You can interact with your container using various services configured by the platform.

🔐 SSH

Use an SSH client (e.g., PuTTY) with:

ssh root@<IP_ADDRESS> -p <PORT>
#  Use the IP, port, and credentials specified during container creation.

📓 Jupyter

Click the Jupyter button under the Description tab to launch Jupyter Notebook.

📈TensorBoard

First, generate data in Jupyter or another environment that produces logs (e.g., training data from machine learning models) then generate TensorBoard

🖥️ CodeServer

Click "Open a Folder" and select "/workspace" to create your project. This is where your project files will be stored and accessed within the container.

---

Delete Container

To delete a container:

• Check the box beside the target container.

• Click the 🗑️ [Delete] icon.

• Confirm using the verification code.

⚠️ Warning: Deleting a container is permanent. Make sure to back up data or store it in a mounted volume.