For AI agents: the complete documentation index is available at https://docs.us.ovhcloud.com/en/llms.txt, the full documentation bundle is available at https://docs.us.ovhcloud.com/en/llms-full.txt, and this page is available as Markdown at https://docs.us.ovhcloud.com/en/public-cloud/ai-machine-learning/manage-registries.md.
  • English

Registries - Use & manage your registries

View as Markdown

Learn how to use and manage your public and private registries through ovhai command line interface (AI CLI), API, and OVHcloud Control Panel. This will help you when working with OVHcloud AI Tools, especially with AI Training and AI Deploy.

 

Requirements

 

Instructions

This documentation is divided into the following parts:

  • Manage existing registries through

    • OVHcloud Control Panel (UI)
    • OVHcloud AI CLI ovhai

  • Create, configure, and use new registries

    • Private Harbor registry
    • Private Docker registry
    • GitHub registry (packages)

 

Manage existing registries

The Public Cloud provides a default registry called shared registry, where users of the same Public Cloud project can push their custom images and use them through our AI Tools.

Warning

NOTE: This shared registry can help you perform your tests but should not be used in production, as we reserve the right to delete its content if deemed necessary. The images pushed to this registry are for AI Tools' workloads only and will not be accessible for external uses.

This is why it can be interesting to add and manage other registries. We can either do this by using the OVHcloud Control Panel (UI) or the ovhai CLI.

Tip

Click each tab across the top to view all content.

Via OVHcloud Control Panel
Via CLI

First, log in to the OVHcloud Control Panel and go to the AI Dashboard section via the Public Cloud menu.

By clicking the Docker registries tab, you should be able to see and manage (add, delete) your different private registries.

Manager AI dashboard docker tab

Note that the shared registry will not appear here. This one is displayed in the Home panel of AI Training.

 

Create, configure, and use new registries

Using your own and private registries is the best way to use your images without exposing them to everyone. There are three main possibilities:

  • Private Harbor registry
  • Private Docker registry
  • GitHub registry / packages
Warning

Port restriction: Only port 443 is supported for registry connections. If you attempt to add a registry URL with a non-443 port (e.g., 8443), the system will enforce port 443. This may lead to errors or connection timeouts if the registry is not configured to accept HTTPS on port 443.

For self-hosted registries using non-443 ports, consider configuring a reverse proxy (e.g., Nginx, HAProxy) to map traffic from 443 to your desired port (e.g., 8443).

We will see how to implement each of these registries and use them through AI Tools.

Info

In any case, note that the default shared registry will remain available even if you add new registries.

Tip

Click each tab across the top to view all content.

Harbor private registry
Docker private registry
GitHub registry

OVHcloud Managed Private Registry service allows you to spawn your own authenticated Docker registry where you can privately store your Docker images. This registry is an instance of Harbor, an open source trusted cloud native registry project that stores, signs, and scans Docker images.

 

Creation

To create your new private registry, we advise you to read and follow the steps mentioned in the Creating a private registry documentation.

At the end of this step, you should have your registry identification details.

 

Configuration

Then, you will need to use these credentials to connect to the Harbor UI. This will allow you to manage this private registry. This simple step is explained in the Connecting to the UI documentation.

To finish setting up your private Harbor registry, you will need to create a project on Harbor, and associate a user to it (Administrator user is added by default, but you can create a new one for this project). You will find all the information you need in this documentation.

 

Get your OVHcloud Managed Private Registry API URL

To add this registry to AI Tools, you will need to retrieve its URL. To do this, go to the Managed Private Registry section on the OVHcloud Public Cloud Manager, click the more options ... button to the right, and choose Harbor API:

PaaS AI general - step 3

Then, copy the URL of the Harbor API, which is the URL of your private registry:

PaaS AI general - step 4

 

Add the Harbor registry

To add your registry, you can either use the OVHcloud Control Panel (UI) or the ovhai CLI.

During this step, you will be asked for your user's credentials (user ID and password). You can use the default user (administrator) credentials, which were obtained during the registry creation step, or those of another user if you have created one.

 

Using UI

To add your private registry via UI, log in to the OVHcloud Control Panel, go to the AI Dashboard section.

From the dashboard, you can add your private Harbor registry by clicking Docker registries > + Add.

Manager AI dashboard docker tab

Here, you will need to provide the credentials of your registry along with its URL.

When you have finished filling in the form, click Add.

Manager AI dashboard docker tab add - step 2

 

Using CLI

To add your private Harbor registry via CLI, use:

ovhai registry add <registry-url>

Once your registry is added, you should see it in your registries list. You can then use any images pushed on this registry for your AI Training jobs and AI Deploy apps.

 

Push an image to your registry

Warning

NOTE: Be careful for the following, we will use the displayed URL. This version does not contain the https:// part, which is present in the URL given by the Public Cloud.

Log in to your private registry, using a user with write rights to your registry project, with the following command:

docker login -u <user> -p <password> <private-registry-address>

Once connected, all you have to do is build, tag, and push your image. Here is an example:

Build the image from the current directory's Dockerfile:

docker build . -t <image-name>:<tag-name>

Tag the compiled image:

docker tag <image-name>:<tag-name> <private-registry-address>/<harbor_project_name>/<image-name>:<tag-name>

Push the image to your private registry's project:

docker push <private-registry-address>/<harbor_project_name>/<image-name>:<tag-name>

 

Use your private registry images with OVHcloud AI Tools

Now that your private registry has been created, added to your Public Cloud account, you can use its image(s) in AI Training and AI Deploy.

Here is a basic example with AI Training, when using the CLI:

ovhai job run <private-registry-address>/<harbor_project_name>/<image-name>:<tag-name>

You can also use the OVHcloud Control Panel if you prefer. In this case, you will have to indicate your custom Docker image during Step 2 of your job/app creation. This path will be <private-registry-address>/<harbor_project_name>/<image-name>:<tag-name>.

Manager AI training order docker

 

Go further

Browse all the AI & Machine Learning guides.

Browse the rest of the OVHcloud US documentation for guides on our other products and services.

Learn more about AI & Machine Learning on the OVHcloud US website.

If you need training or technical assistance to implement our solutions, contact your sales representative or click on this link to get a quote and ask our Professional Services experts for a custom analysis of your project.