Container Backup & Recovery

The Problem with Full Backups

A typical container image is hundreds of megabytes: most of it unchanged base layers (OS, runtime, dependencies). Backing up the entire filesystem every time is slow, expensive, and wasteful.

reel solves this by capturing just the filesystem layer: everything that changed since the container started (config files, logs, application data).

What You Can Do

• Backup - Regular filesystem snapshots in seconds, local or remote
• Recovery - Restore a known-good state to a new pod
• Rollback - Revert current deployment pods to previous layer
• Forensics - Store and preserve evidence of file modifications

Find Your Workloads

List pods that reel can operate on:

Export Layer

Captures the current filesystem state and exports it directly to S3 or a local file. No copy is left on the node.

See S3 Exports for bucket configuration.

Create a Layer

Captures the current filesystem state and saves it on the node where the pod is running.

You can specify the container name if the pod has multiple containers.

Output:

List Local Layers

Restore to New Pod

Create a standalone pod from a layer. Use this for debugging or forensic analysis without affecting the live deployment.

This creates recovered-api with the filesystem state from the layer.

Rollback Deployment

Revert a live deployment to a previous layer. Use this to recover from bad updates or configuration changes.

Automated Local Backups

Add an annotation for scheduled captures with built-in rotation:

See Scheduling for full configuration.

Retention Management

Local layers are stored on each node and consume disk space. Use retention policies to automatically clean up old layers and keep storage under control.

You can combine retention with scheduled backups using the next keyword, as shown in the automated backups example above.