Deploys.app Docs

Disks

A disk is a managed block volume that survives container restarts and rollouts. Create it once, attach it to a deployment, and the data stays put across every redeploy.

The Disks page#

The Disks tab lists every disk in the project, with its size in GiB and the location it lives in.

console.deploys.app/disk?project=acme
Disks: postgres-data 50 GiB, uploads 100 GiB, redis-data 10 GiB
Three disks — each tied to a single location, sized for the workload it backs.

Create a disk#

You’ll set three things: a name (must be unique within the project + location), a size in GiB, and the location. The location matters: a disk can only be attached to deployments in the same location.

deploys disk create \
  --project acme --location gke.cluster-rcf2 \
  --name uploads --size 100

The size is the initial size. You can grow it later — see Resize a disk below — but you can’t shrink it.

Attach a disk to a deployment#

Mount a disk by adding disk to the deployment’s configuration. The disk must exist and be in the same (project, location), and must not already be attached to another deployment.

{
  "name": "api",
  "image": "registry.deploys.app/acme/api:v1.8.0",
  "disk": { "name": "uploads", "mountPath": "/data" }
}

mountPath is the absolute path inside the container. Optional subPath mounts a sub-directory of the disk only — useful if you want one disk to back multiple deployments, each writing into its own folder.

{
  "disk": { "name": "shared-data", "mountPath": "/data", "subPath": "api" }
}
WarningOnly one deployment can mount a given disk at a time. Attempts to attach an already-mounted disk fail at deploy time. If you need shared storage across multiple deployments, you currently need a separate disk per deployment (or mount the same disk with different subPaths, with the understanding that each deployment writes only to its own subpath).

Resize a disk#

Disks can grow. The new size must be larger than the current size.

deploys disk update \
  --project acme --location gke.cluster-rcf2 \
  --name uploads --size 200

The platform expands the underlying volume; the deployment that owns it doesn’t need to restart, but the filesystem inside the container may not see the new space until the container is restarted or runs resize2fs (depending on the filesystem).

Delete a disk#

deploys disk delete \
  --project acme --location gke.cluster-rcf2 \
  --name old-data

Deletion fails if the disk is currently attached. Detach it first by re-deploying without disk (or by deleting the deployment).

What disks are good for#

  • Databases you self-host — Postgres, MySQL, Redis with persistence on.
  • Uploaded user content — until you move it to object storage.
  • Local caches you want to survive restarts — search indexes, image thumbnails, etc.

What they’re not good for: anything you’d reach for shared object storage for. A disk is bound to one location and one deployment at a time. For multi-region or multi-deployment access, push data to a bucket and serve it from there.

On this page