Cluster & Host Backups

Cluster & Host Backups backs up the actual infrastructure a connected agent monitors: the Kubernetes objects in a cluster, or the named volumes and bind mounts on a Docker host. This is available on the Pro and Enterprise plans.

What it does

For a Kubernetes agent, on the schedule you configure:

1. Exports every object in the namespaces you choose (or your whole cluster) as manifests, filtered by a label selector if you set one.
2. Bundles each namespace's manifests into an archive.
3. Pushes the archive to every destination you've configured.
4. Records the result per namespace.

For a Docker agent, the same schedule instead archives the named volumes and bind mounts you list, and pushes each one to your configured destinations.

You can also trigger a backup immediately without waiting for the schedule.

Enabling it

Each agent has to opt in before it will accept a backup command. This is off by default, separate from the read-only access an agent gets on install.

Kubernetes, add this to your helm upgrade/helm install:

--set rbac.allowClusterBackup=true

Docker, set this environment variable on the agent container:

AGENT_ALLOW_BACKUP=true

Once the agent has the flag, open System Health in the sidebar, find the agent in Connected Agents, and click Backups. If your plan doesn't include this feature, you'll see an upgrade prompt instead. See Plans & Pricing.

Choosing what gets backed up

Kubernetes: leave namespaces blank to back up the whole cluster, or list which namespaces to include or exclude, and optionally a label selector to narrow it further within that set.

Docker: list the containers and mounts (named volumes or bind mounts) you want covered. There's no "everything" default for Docker, since there's no equivalent of a namespace to scope by.

Destinations, scheduling, and retention

Add at least one destination (S3, SFTP, webhook, or Azure Blob), the same four kinds as Disaster Recovery. Credentials are entered once and never shown again in the UI. Bytes are pushed to your destination directly from the agent, never through KubeWatch's own servers.

Turn on automatic backups and set the interval (24 hours by default) and retention (30 days by default). You can also click Run now to trigger a backup immediately, independent of the schedule.

Viewing backup history

The same page lists every backup that's run: when, how it was triggered, and a per-item breakdown (which namespaces or mounts, and whether each one succeeded).

Limitations (this release)

• This backs up Kubernetes object manifests and Docker volume/bind-mount data. It does not yet capture Kubernetes PersistentVolume data itself (a future release adds this via CSI volume snapshots).
• There's no restore or migrate action yet. Today this produces a portable copy of your manifests and volumes at a destination you control; a one-click restore into the same or a different cluster is planned for a future release.
• There's no pre- or post-backup hook support yet (for example, flushing a database to disk before its volume is backed up).