> ## Documentation Index
> Fetch the complete documentation index at: https://docs.scitix.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Container Image Registry Overview

> Learn about the use cases, core objects, and usage workflow of the container registry.

Container Registry is a private repository in GPU Cloud for storing, managing, and distributing container images and Helm Charts. You can push locally built images or Charts to projects in a specified region, or import images from external repositories for use in local environments, CI/CD pipelines, or CKS clusters.

## Applicable Scenarios

Container image registries are suitable for the following scenarios:

* Before deploying containerized workloads in a CKS cluster, centrally store workload images.
* Manage team-built images by project and tag for pulling in development, testing, or production environments.
* Store Helm Charts to reuse Kubernetes application deployment configurations.
* Import images from external repositories into GPU Cloud, or synchronize images from GPU Cloud to other regions or external repositories.
* Use audit logs, webhooks, and retention policies to trace actions, integrate with downstream systems, and clean up historical artifacts.

## Core Concepts

### Region and Repository Address

The registry provides services by region, and each region has a different registry address. When pushing, pulling, or synchronizing images, use the registry address corresponding to the region where the image is located or the target region.

### Project

A project is used to organize a group of container images or Helm Charts and is also part of an image path. Before pushing, confirm the target region and project. When pulling, you typically need the repository address, project name, image or Chart name, and version information.

Projects can be configured with members and access scopes. When business images, internal components, or other sensitive content are involved, we recommend confirming the access scope when creating the project.

### Artifacts

Container images and Helm Charts are both artifacts in the registry. You can push and pull container images using Docker, and push and pull Charts using Helm. When you need to clean up historical versions, you can configure an artifact retention policy.

### Copy Rules

Replication rules are used to transfer images between different repositories. Select a mode based on the image flow direction:

* Use Pull Mode to import images from an external repository into the GPU Cloud.
* Use Push Mode to synchronize images in the GPU Cloud to other regions or external repositories.

### Governance and Tracking

The registry provides retention policies, webhooks, and audit logs.

* Retention policies are used to clean up historical artifacts that are no longer needed.
* Webhooks are used to notify downstream systems when artifacts change.
* Audit logs are used to query action records in the Registry.

## Usage workflow

**1. Prepare a Project and Members**

Before pushing a Docker image or Helm Chart, create a project in the target region. A project is part of the image path and is also used to manage the access scope of artifacts.

If team members, CI/CD pipelines, or automated tasks need to access artifacts in the project, add members to the project and configure permissions. For details, see [Manage Registry Projects and Members](https://docs.scitix.ai/gpu-cloud/registry/create-view-container-registry-projects.md).

**2. Push and Pull Artifacts**

Select the appropriate push and pull method based on the artifact type.

* Docker images: Use Docker to log in to the repository and push local images to the project, or pull and use them in local environments, CI/CD pipelines, or CKS clusters. For details, see [Push and Pull Container Images](https://docs.scitix.ai/gpu-cloud/registry/push-pull-container-images.md).
* Helm Charts: Use Helm to log in to the repository and push Charts to the project by version, or pull a specified version in other environments. For details, see [Push and Pull Helm Charts](https://docs.scitix.ai/gpu-cloud/registry/push-pull-helm-charts.md).

**3. Copy Image**

If the image needs to be transferred between different repositories or regions, select a copy method based on the image flow.

* When importing an image from an external repository, use Pull Mode. For details, see [Importing Images from External Repositories](https://docs.scitix.ai/gpu-cloud/registry/import-external-container-images.md).
* When synchronizing an image to another region or an external repository, use Push Mode. For details, see [Synchronizing Images to Other Regions or Repositories](https://docs.scitix.ai/gpu-cloud/registry/sync-container-images.md).

**4. Govern and Track Artifacts**

After an image is added to a project, you can configure governance capabilities based on retention, integration, and audit requirements.

* To clean up unused artifacts, configure the [Registry Artifact Retention Policy](https://docs.scitix.ai/gpu-cloud/registry/configure-artifact-retention.md).
* To notify downstream systems when artifacts change, configure the [Registry Webhook](https://docs.scitix.ai/gpu-cloud/registry/configure-container-registry-webhook.md).
* To trace actions related to projects, artifacts, or replication, view the [Registry Audit Logs](https://docs.scitix.ai/gpu-cloud/registry/view-container-registry-audit-logs.md).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.