Rediacc CLI Reference

Overview

The **Rediacc CLI** (`rdc`) is a command-line interface for managing self-hosted infrastructure — machines, repositories, storage, backups, and more.

Installation

Download the latest binary for your platform from the [Rediacc releases page](https://www.rediacc.com) or install via your package manager:

# macOS / Linux
curl -fsSL https://www.rediacc.com | sh

# Or use the packaged binary directly
./rdc --help

Global Options

Every command supports these global flags:

Flag	Description
`--output`	Output format (table\|json\|yaml\|csv). Override default with REDIACC_DEFAULT_OUTPUT.
`--config`	Use specific config for this command
`--lang`	Language override ({{languages}})
`--force`	Skip confirmation prompts

1. Config Management

Manage configuration files: create, switch, edit, reconcile state, and cross-cutting facilities (fields, audit, SSH keys, remote sync).

Config files store connection details (API URL, credentials, default team/region/bridge) so you can switch between environments quickly.

1.1 prune

Remove dead weight from the local config file at ~/.config/rediacc/<config>.json. Three buckets are cleaned, all pure-local (no SSH/renet calls): (1) ACME cert-cache entries whose anchor GUID/repo/machine is no longer in the active config; (2) archived repositories whose grace period has expired (default 7 days, see defaults.pruneGraceDays); (3) dangling cross-references (machine→strategy, strategy→repo). Resources still in use, credentials, storage tokens, and known-hosts are never touched. Default behavior is to apply changes; pass --dry-run to preview only.

rdc config prune [options]

Flag	Description	Required	Default
`--dry-run`	Preview what would be removed without modifying the config file. Mirrors the default-off semantics of the other prune commands.	No	-
`--certs-only`	Restrict to the ACME cert-cache bucket. Skips archive purging and cross-reference cleanup. Mutually exclusive with --archives-only and --refs-only.	No	-
`--archives-only`	Restrict to expired-archive purging. Skips cert-cache and cross-reference cleanup. Mutually exclusive with --certs-only and --refs-only.	No	-
`--refs-only`	Restrict to dangling cross-references (machine→strategy, strategy→repo excludes/includes). Skips cert-cache and archives. Mutually exclusive with --certs-only and --archives-only.	No	-
`--purge-archived`	Drop ALL archived repositories regardless of age, not just those past grace. Equivalent to running 'rdc repo admin archive purge' for every archived repo. Use only when you're sure you don't need any of the stashed credentials for restore.	No	-
`--orphan-repos`	Remove repository entries that are placed on no machine. Each entry holds that repo's credential and SSH key, so this is unrecoverable — run 'rdc config reconcile' first so placement reflects the machines.	No	-
`--grace-days <days>`	Override the archive grace window (in days) for this invocation. Falls back to defaults.pruneGraceDays in the config, then to 7 if neither is set.	No	-

1.2 init

Create a new named config file

You can add machines directly:

rdc config init [name] [options]

Flag	Description	Required	Default
`--ssh-key <path>`	Path to SSH private key (e.g., ~/.ssh/id_rsa)	No	-
`--renet-path <path>`	Path to renet binary (default: renet in PATH)	No	-
`--master-password <password>`	Encrypt resources with a master password	No	-
`--server <url>`	Account server URL	No	-

1.3 list

List all config files

rdc config list

CLI reference: rdc config list

1.4 show

Show current config details

rdc config show [options]

Flag	Description	Required	Default
`--reveal`	Show plaintext for sensitive values (interactive only)	No	-

1.5 current

Show the active config: name, file, resolved server and channel, token state

rdc config current

CLI reference: rdc config current

1.6 delete

Delete a config file

rdc config delete <name>

CLI reference: rdc config delete

1.7 set

Set a default value (team, region, bridge)

rdc config set <key> <value>

CLI reference: rdc config set

1.8 clear

Clear defaults (all or specific key)

rdc config clear [key]

CLI reference: rdc config clear

1.9 recover

Restore config from backup (.bak) file

rdc config recover [name] [options]

Flag	Description	Required	Default
`-y, --yes`	Skip confirmation prompt	No	-

1.10 ssh

Manage SSH key configuration

set

Set SSH key for the current config

rdc config ssh set [options]

Flag	Description	Required	Default
`--key <path>`	Path to SSH private key file	Yes	-
`--embed`	Embed key content in config instead of storing path	No	-

show

Show current SSH key configuration

rdc config ssh show

CLI reference: rdc config ssh show

remove

Remove SSH key from the current config

rdc config ssh remove

CLI reference: rdc config ssh remove

1.11 remote

Manage remote config storage connection

enable

Link this config to remote encrypted storage

rdc config remote enable [options]

Flag	Description	Required	Default
`--headless`	Use device code flow (for headless servers)	No	-
`--password`	Enroll headlessly with a pre-provisioned password slot (no browser)	No	-
`--api-url <url>`	Account server URL	No	-
`--force`	Replace differing local content with the server copy without confirmation	No	-

disable

Disconnect from remote storage and save config locally

rdc config remote disable

CLI reference: rdc config remote disable

status

Show remote connection status

rdc config remote status

CLI reference: rdc config remote status

refresh

Force re-fetch config from remote storage

rdc config remote refresh

CLI reference: rdc config remote refresh

1.12 field

Pointer-addressed config field operations (get/set/unset/rotate/list). JSON Pointers (RFC 6901) like /credentials/cfDnsApiToken.

get

Read a single config value by JSON Pointer. Sensitive fields redact unless --reveal (humans only).

rdc config field get [options]

Flag	Description	Required	Default
`--pointer <pointer>`	JSON Pointer to the field (e.g. /credentials/cfDnsApiToken)	Yes	-
`--reveal`	Show plaintext for sensitive values (interactive TTY only; audited)	No	-
`--digest`	Print the SHA-256 digest instead of the value (safe to share with agents)	No	-

set

Write a config value at a JSON Pointer. Sensitive paths require --current (knowledge-gate).

rdc config field set [options]

Flag	Description	Required	Default
`--pointer <pointer>`	JSON Pointer to the field (e.g. /credentials/cfDnsApiToken)	Yes	-
`--new <value>`	New value (parsed as JSON if it looks like JSON: {, [, ", true/false/null/number)	Yes	-
`--current <value>`	Current plaintext value — required for sensitive-path mutations (knowledge-gate proof)	No	-

unset

Delete a config value at a JSON Pointer. Sensitive paths require --current.

rdc config field unset [options]

Flag	Description	Required	Default
`--pointer <pointer>`	JSON Pointer to the field (e.g. /credentials/cfDnsApiToken)	Yes	-
`--current <value>`	Current plaintext value — required for sensitive-path deletions	No	-

rotate

Rotate a sensitive value without --current. Interactive TTY only; loudly audited.

rdc config field rotate [options]

Flag	Description	Required	Default
`--pointer <pointer>`	JSON Pointer to the sensitive field (e.g. /credentials/cfDnsApiToken)	Yes	-
`--new <value>`	New value	Yes	-

list

List every registered sensitivity pointer template with its kind and commit/encrypt policy.

rdc config field list [options]

Flag	Description	Required	Default
`--sensitive`	Show only sensitive (non-public) templates	No	-

1.13 edit

Open the active config in $EDITOR as a redacted JSONC projection. Humans only; agents refused.

rdc config edit [options]

Flag	Description	Required	Default
`--reveal`	Show plaintext for sensitive values (interactive TTY only; audited)	No	-
`--dump`	Print current config as JSONC to stdout (read-only; safe for agents when redacted)	No	-
`--apply <file>`	Apply an edited JSONC file (skips $EDITOR launch)	No	-
`--current-secrets <file>`	JSON file mapping pointer→old plaintext for knowledge-gate on --apply	No	-
`--editor <cmd>`	Editor command override (follows git precedence: flag > $GIT_EDITOR > git config core.editor > $VISUAL > $EDITOR)	No	-

1.14 audit

Inspect the config audit log (hash-chained JSONL at ~/.config/rediacc/audit.log.jsonl)

log

Print recent audit entries as JSON

rdc config audit log [options]

Flag	Description	Required	Default
`--since <spec>`	Only show entries newer than (e.g., '24h', '7d', ISO timestamp)	No	-
`--path <glob>`	Filter by JSON Pointer glob (e.g., /credentials/*)	No	-
`--actor <kind>`	Filter by actor kind (human\|agent)	No	-

tail

Stream new audit entries as they are written (Ctrl+C to stop)

rdc config audit tail

CLI reference: rdc config audit tail

verify

Verify the integrity of the SHA-256 hash chain across all audit entries

rdc config audit verify

CLI reference: rdc config audit verify

1.15 reconcile

Rebuild runtime state from machine truth. Fixes stale attach and routing data.

rdc config reconcile [options]

Flag	Description	Required	Default
`--machine <m...>`	Limit the reconcile to these machines	No	-
`--dry-run`	Show what would be done without making changes	No	-
`--accept-observed`	Rewrite a declared placement to match where the image actually is, but only when that is unambiguous (observed on exactly one machine). Duplicates stay conflicts.	No	-

1.16 rotate-cek

Rotate the organization config-encryption key

rdc config rotate-cek [options]

Flag	Description	Required	Default
`--api-url <url>`	Account server URL	No	-

2. Datastore Management

Named storage pools that hold repositories. A datastore is mobile (attach it to a machine, move it to another) and single-mounter (exactly one machine holds it at a time). The implicit default datastore on each machine is not managed here.

2.1 resize

Grow or shrink a datastore. This is an offline operation: the repositories inside it must be stopped.

rdc datastore resize <datastore> [options]

Flag	Description	Required	Default
`--size <size>`	New size (for example 200G)	Yes	-
`--debug`	Enable debug output	No	-

2.2 fork

Fork a datastore copy-on-write. The fork is instant and its cost does not grow with the size of the pool. Needs the rbd backend: a local datastore has no block-level clone, so repositories inside it fork one at a time by reflink instead (rdc repo fork).

rdc datastore fork <datastore> [options]

Flag	Description	Required	Default
`--tag <tag>`	Tag for the fork (the result is name:tag)	Yes	-
`--attach-to <machine>`	Attach the fork to this machine right away (needs --writes)	No	-
`--writes <disposition>`	Where the fork's writes go: local (instant, ephemeral) or ceph (durable clone)	No	-
`--cow-size <size>`	Overlay size for --writes local	No	-
`--debug`	Enable debug output	No	-

2.3 status

Show one datastore: its backend, attachment, usage, repositories and snapshots. A detached datastore still reports its record.

rdc datastore status <datastore> [options]

Flag	Description	Required	Default
`--debug`	Enable debug output	No	-

2.4 create

Create an additional named datastore on a machine. A local backend is a file-backed pool on that machine and stays there. An rbd backend lives in Ceph, so it can move to any machine that reaches the same cluster. Pass --cluster to make it a kubernetes-world datastore; that backref is fixed at creation.

rdc datastore create <datastore> [options]

Flag	Description	Required	Default
`-m, --machine <name>`	Machine to create the datastore on	Yes	-
`--size <size>`	Datastore size (for example 100G)	Yes	-
`--backend <type>`	Storage backend: local (file-backed, stays on this machine) or rbd (Ceph, movable)	No	`local`
`--pool <name>`	Ceph pool for the rbd backend (default: rbd)	No	-
`--image <name>`	RBD image name for the rbd backend (default: the datastore name)	No	-
`--cluster <name>`	Kubernetes cluster this datastore belongs to. Set means kubernetes repositories only; unset means docker repositories only. Fixed at creation.	No	-
`--debug`	Enable debug output	No	-

2.5 list

List named datastores, where they are attached, and what they hold.

rdc datastore list [place]

CLI reference: rdc datastore list

2.6 attach

Attach a datastore to a machine. Exactly one machine holds a datastore at a time, so attaching it somewhere else moves it: the old holder gives it up first, and a failed hand-off leaves the original attachment intact. A fork must say where its writes go.

rdc datastore attach <datastore> [options]

Flag	Description	Required	Default
`--to <machine>`	Machine to attach the datastore to	Yes	-
`--writes <disposition>`	Where a fork's writes go: local (instant, ephemeral, lost on detach) or ceph (durable clone in the pool). Required for a fork.	No	-
`--cow-size <size>`	Overlay size for --writes local	No	-
`--no-auto`	Do not re-attach this datastore automatically on boot	No	-
`--force`	Fence a stale holder that did not give the datastore up cleanly	No	-
`--debug`	Enable debug output	No	-

2.7 detach

Detach a datastore from its machine. Repositories inside it stop first. A fork attached with --writes local has nowhere to write its overlay back to, so detaching it throws the overlay away and needs --discard.

rdc datastore detach <datastore> [options]

Flag	Description	Required	Default
`--discard`	Throw away a local-writes fork and its overlay. The data is not recoverable.	No	-
`-y, --yes`	Skip confirmation prompt	No	-
`--debug`	Enable debug output	No	-

2.8 snapshot

Take and list point-in-time snapshots of a datastore. A snapshot costs nothing at rest and is what a fork clones from.

create

Take a point-in-time snapshot of a datastore. Nothing stops. The instant is crash-consistent, not application-consistent: if you need a specific write to be IN the snapshot, sync it first.

rdc datastore snapshot create <datastore> [options]

Flag	Description	Required	Default
`--snapshot <label>`	Snapshot label (default: a UTC timestamp)	No	-
`--debug`	Enable debug output	No	-

list

List a datastore's snapshots.

rdc datastore snapshot list <datastore> [options]

Flag	Description	Required	Default
`--debug`	Enable debug output	No	-

2.9 delete

Destroy a datastore and everything in it. It detaches first; if it will not detach cleanly, the delete fails rather than orphaning a mounted pool.

rdc datastore delete <datastore> [options]

Flag	Description	Required	Default
`-y, --yes`	Skip confirmation prompt	No	-
`--force`	Delete even though repositories still point at it. Their data goes with it.	No	-
`--debug`	Enable debug output	No	-

3. Machine Management

Inspect remote machine state: system resources, running containers, systemd services, deployed repositories, and health diagnostics. Read-only queries; to modify repos use `repo`. Key subcommands: status (full state, narrowed with --system, --containers, --repositories, --services), health, list, add, remove, setup, prune, provision, deprovision.

3.1 list

List machines

rdc machine list [options]

Flag	Description	Required	Default
`--search <text>`	Search in {{field}}	No	-
`--sort <field>`	Sort by field	No	-
`--desc`	Sort in descending order	No	-

3.2 health

Check machine health for CI/CD pipelines

The health command outputs system metrics, container status, service stability, and storage information — useful for CI/CD pipeline checks.

rdc machine health <name> [options]

Flag	Description	Required	Default
`-t, --team <name>`	Team name	No	-

3.3 provision

Provision a new machine on a cloud provider using OpenTofu

rdc machine provision <name> [options]

Flag	Description	Required	Default
`--provider <name>`	Cloud provider name (from machine provider add)	Yes	-
`--region <region>`	Override default region	No	-
`--type <type>`	Override default instance type	No	-
`--image <image>`	Override default OS image	No	-
`--ssh-user <user>`	SSH username for the new VM (default: root)	No	-
`--base-domain <domain>`	Base domain for infrastructure (e.g., example.com). Implies --infra	No	-
`--no-infra`	Skip infrastructure configuration (proxy + DNS)	No	-
`--debug`	Enable debug output	No	-

3.4 deprovision

Destroy a cloud-provisioned machine and remove from config

rdc machine deprovision <name> [options]

Flag	Description	Required	Default
`--force`	Skip confirmation prompt	No	-
`--debug`	Enable debug output	No	-

3.5 prune

Remove orphaned datastore resources and stale snapshots from a machine. The base run cleans renet-internal datastore artifacts (BTRFS subvolumes, lock files, tmpfiles). The optional flags below enable progressively narrower repo cleanups: --orphaned-repos uses the local CLI config as the only signal, while --prune-unknown additionally consults the renet .interim/state mirror so legitimate forks created by other tools survive even when missing from your local config. Both deletion paths run a mount-safety preflight; pass --force-delete-mounted to override.

rdc machine prune <name> [options]

Flag	Description	Required	Default
`--dry-run`	Show what would be removed without making changes	No	-
`--orphaned-repos`	Delete every repo image on the machine that is not in your local CLI config. Coarse — also removes forks created by other tools that have no local config entry, even when their renet mirror correctly identifies them as forks. Use --prune-unknown for the narrower behavior that respects the mirror.	No	-
`--prune-unknown`	Delete only repos the renet .interim/state mirror cannot classify (not in local config AND no fork-marked mirror). Strictly narrower than --orphaned-repos: forks-without-config are preserved when the mirror identifies them. Pre-mirror legacy orphans and stale grands whose config entry was deleted both fall in this bucket.	No	-
`--force-delete-mounted`	Override the mount-safety preflight and delete repos even if they are currently mounted or have running Docker containers. Distinct from --force (which only overrides the archive grace period). Applies to both --orphaned-repos and --prune-unknown.	No	-
`--force`	Skip confirmation prompts	No	-
`--grace-days <days>`	Grace period in days for recently archived repos (default: 7)	No	-
`--debug`	Enable debug output	No	-

3.6 add

rdc machine add <name> [options]

Flag	Description	Required	Default
`--ip <address>`	Machine IP address or hostname	Yes	-
`--user <username>`	SSH username	Yes	-
`--port <port>`	SSH port	No	`22`

3.7 infra

Manage infrastructure configuration (proxy, DNS, certificates)

cert

Manage cached TLS certificates

clear:

Remove the certificate cache

rdc machine infra cert clear

CLI reference: rdc machine infra cert clear

pull:

Download and cache TLS certificates from a machine

rdc machine infra cert pull <machine> [options]

Flag	Description	Required	Default
`--no-prune`	Skip pruning stale network-ID certificates	No	-
`--debug`	Enable debug output	No	-

push:

Upload cached TLS certificates to a machine

rdc machine infra cert push <machine> [options]

Flag	Description	Required	Default
`--debug`	Enable debug output	No	-

status:

Show cached certificate inventory

rdc machine infra cert status

CLI reference: rdc machine infra cert status

push

Push infrastructure config to machine (Traefik proxy, router, Cloudflare DNS). Run 'machine infra set <machine>' first

rdc machine infra push <machine> [options]

Flag	Description	Required	Default
`--debug`	Enable debug output	No	-

set

Set infrastructure configuration for a machine (machine-specific: IPs, domain, ports; shared: cert email, CF DNS token)

rdc machine infra set <machine> [options]

Flag	Description	Required	Default
`--public-ipv4 <ip>`	Public IPv4 address (per-machine)	No	-
`--public-ipv6 <ip>`	Public IPv6 address (per-machine)	No	-
`--base-domain <domain>`	Base domain for applications (per-machine)	No	-
`--cert-email <email>`	Email for TLS certificate notifications (shared across machines)	No	-
`--cf-dns-token <token>`	Cloudflare DNS API token for ACME DNS-01 challenge (shared across machines)	No	-
`--tcp-ports <ports>`	TCP ports to forward (comma-separated, e.g., 25,143,465)	No	-
`--udp-ports <ports>`	UDP ports to forward (comma-separated, e.g., 53)	No	-

show

Show infrastructure configuration for a machine

rdc machine infra show <machine>

CLI reference: rdc machine infra show

3.8 provider

Manage cloud providers for automated machine provisioning

add

Add a cloud provider

rdc machine provider add <name> [options]

Flag	Description	Required	Default
`--provider <source>`	Known provider source (e.g., linode/linode, hetznercloud/hcloud)	No	-
`--source <source>`	Custom OpenTofu provider source (e.g., vultr/vultr)	No	-
`--token <token>`	API token for the cloud provider	Yes	-
`--region <region>`	Default region for new machines	No	-
`--type <type>`	Default instance type/size	No	-
`--image <image>`	Default OS image	No	-
`--ssh-user <user>`	SSH username for new VMs (default: root)	No	-
`--resource <type>`	Custom: OpenTofu resource type for VM	No	-
`--label-attr <attr>`	Custom: attribute name for VM label	No	-
`--region-attr <attr>`	Custom: attribute name for region	No	-
`--size-attr <attr>`	Custom: attribute name for instance type	No	-
`--image-attr <attr>`	Custom: attribute name for OS image	No	-
`--ipv4-output <attr>`	Custom: output attribute for IPv4 address	No	-
`--ipv6-output <attr>`	Custom: output attribute for IPv6 address	No	-
`--ssh-key-attr <attr>`	Custom: attribute name for SSH keys	No	-
`--ssh-key-format <format>`	Custom: SSH key format (inline_list or resource_id)	No	-
`--ssh-key-resource <type>`	Custom: OpenTofu resource type for SSH keys	No	-

list

List configured cloud providers

rdc machine provider list

CLI reference: rdc machine provider list

remove

Remove a cloud provider configuration

rdc machine provider remove <name>

CLI reference: rdc machine provider remove

3.9 remove

Remove a machine from the config. Does not touch the machine itself.

rdc machine remove <name> [options]

Flag	Description	Required	Default
`-y, --yes`	Skip confirmation prompt	No	-
`--force`	Remove the machine even if repositories are still placed on it, leaving those placements dangling.	No	-

3.10 scan-keys

Scan SSH host keys for machines in the current config

rdc machine scan-keys [name]

CLI reference: rdc machine scan-keys

3.11 setup

Install renet and prepare a registered machine for repositories.

rdc machine setup <name> [options]

Flag	Description	Required	Default
`--datastore-path <path>`	Datastore path on remote machine	No	`/mnt/rediacc`
`--datastore-size <size>`	Datastore size (e.g., 95%, 100G)	No	`95%`
`--debug`	Enable debug output	No	-

3.12 status

Show a machine's system, repositories, containers, and services.

rdc machine status [name] [options]

Flag	Description	Required	Default
`--debug`	Enable debug output	No	-
`--system`	Include system info only	No	-
`--repositories`	Include repositories only	No	-
`--containers`	Include containers only	No	-
`--services`	Include services only	No	-
`--network`	Include network interfaces only	No	-
`--block-devices`	Include block devices only	No	-
`--licenses`	Include repository license statuses	No	-
`--storage-health`	Show BTRFS reflink savings and (informational) image fragmentation per repository	No	-
`--datastores`	Show attached datastores (mount, attach, usage)	No	-
`--health-check`	Health check mode - exits with code 2 if any unhealthy	No	-
`--stability-check`	Stability check mode - exits with code 2 if any failed/restarting	No	-
`--search <text>`	Filter repositories by name	No	-
`--sync-certs`	Also pull the ACME cert cache from the machine after querying	No	-
`--strict`	Exit non-zero (code 2) if any container has crossed the health-drift threshold	No	-

4. Cluster Management

Provision and manage clusters: named sets of node pools (ceph, k8s-server, k8s-agent, hyperconverged) on a private network. Pool members materialize into machines, so every -m command works on them.

4.1 create

Declare and provision a cluster: machines, Ceph pools, and Kubernetes. Pass --provider and --pool to declare in one step; a bare create provisions a cluster already declared.

rdc cluster create <cluster> [options]

Flag	Description	Required	Default
`--provider <provider>`	Provider: a cloudProviders key, or 'kvm'	No	-
`--pool <spec...>`	Pool spec name:role:count[:size] (role: ceph\|k8s-server\|k8s-agent\|hyperconverged)	No	-
`--declare-only`	Record the cluster in config without provisioning it	No	-
`--network-cidr <cidr>`	Private network CIDR (e.g. 10.0.0.0/24)	No	-
`--network-primitive <primitive>`	Network primitive (e.g. vlan, vpc, network)	No	-
`--control-node <machine>`	Explicit control-node machine (default: first k8s-server member)	No	-
`--net-name <name>`	KVM: libvirt network for this cluster (e.g. renet12)	No	-
`--net-base <prefix>`	KVM: network prefix, the first three octets (e.g. 192.168.112)	No	-
`--net-offset <n>`	KVM: offset added to each VM id when deriving its address	No	-
`--control-id <n>`	KVM: VM id of the control and registry node (default: 1)	No	-
`--docker-registry <endpoint>`	KVM: in-VM Docker registry endpoint for this cluster	No	-
`--ssh-user <user>`	SSH user for provisioned members	No	-
`--base-domain <domain>`	Base domain for cluster public DNS (else inherited from a sibling machine)	No	-
`--control-ds-size <size>`	Anchor control datastore size (default 10G)	No	-
`--control-ds-backend <backend>`	Anchor control datastore backend: local \| ceph (default: ceph if the cluster has ceph, else local)	No	-
`--control-ds-pool <pool>`	Ceph rbd pool for the anchor control datastore (ceph backend)	No	-
`--debug`	Enable debug output	No	-

4.2 status

List all clusters, or pass a cluster name to show that cluster's full config.

rdc cluster status [cluster]

CLI reference: rdc cluster status

4.3 scale

Change a pool's member count (adds/removes machines and joins/drains nodes). k8s-agent pools scale in place; ceph pool growth is done via cephadm directly.

rdc cluster scale <cluster> [options]

Flag	Description	Required	Default
`--pool <pool>`	Pool name	Yes	-
`--count <n>`	Desired member count	Yes	-
`--debug`	Enable debug output	No	-

4.4 destroy

Tear down the provisioned members and remove the cluster and its machines from config.

rdc cluster destroy <cluster> [options]

Flag	Description	Required	Default
`--force`	Skip confirmation and continue past teardown errors	No	-
`--debug`	Enable debug output	No	-

4.5 kubeconfig

Fetch the kubeconfig from the cluster control node over SSH and cache it locally (0600) for kubectl.

rdc cluster kubeconfig <cluster>

CLI reference: rdc cluster kubeconfig

4.6 fork

Clone an entire cluster, including its repos' data, into a new cluster: coordinated copy-on-write of the cluster and PV images, then node-identity rewrite so the fork comes up on new addresses. The parent keeps running.

rdc cluster fork <cluster> [options]

Flag	Description	Required	Default
`--tag <tag>`	Fork tag	Yes	-
`--to <dest-cluster>`	Destination cluster whose nodes host the fork (its control node and agents; a cluster cannot fork onto its own machines, since two k3s cannot share a host network namespace)	Yes	-
`--writes <disposition>`	Fork write disposition: local (ephemeral dm-COW overlay, zero Ceph footprint) \| ceph (durable clone). Default local	No	-
`--up`	Bring the forked repos up and gate on cluster health after the fork boots	No	-
`--debug`	Enable debug output	No	-

4.7 migrate

Move an entire cluster, including its repos' data, to another machine or datacenter with a short cutover (hot pre-copy, then stop-and-restart on the destination).

rdc cluster migrate <cluster> [options]

Flag	Description	Required	Default
`--to <dest-cluster>`	Destination	Yes	-
`--debug`	Enable debug output	No	-

4.8 join

Adopt an existing registered machine as a Kubernetes agent node of a cluster, using the same CA-derived join token as anchor and rejoin.

rdc cluster join <machine> [options]

Flag	Description	Required	Default
`--cluster <name>`	Cluster to join the machine to	Yes	-
`--debug`	Enable debug output	No	-

4.9 evict

Drain the node, delete its Node object, and clear its cluster membership. The cluster is derived from the machine; a machine that still mounts a datastore is refused.

rdc cluster evict <machine> [options]

Flag	Description	Required	Default
`--force`	Skip the drain when the node is already dead	No	-
`--debug`	Enable debug output	No	-

4.10 rehearse

Rehearse a release/upgrade: fork the cluster onto a destination as an ephemeral throwaway (writes=local, secretless role=rehearsal), bring it up, gate on health, then discard it. The parent is never touched.

rdc cluster rehearse <cluster> [options]

Flag	Description	Required	Default
`--on <dest-cluster>`	Destination cluster to boot the throwaway rehearsal on	Yes	-
`--tag <tag>`	Optional tag for the rehearsal fork (default: timestamped)	No	-
`--debug`	Enable debug output	No	-

4.11 snapshot

Point-in-time snapshots of the whole cluster. One crash-consistent instant across every rbd-backed datastore the cluster owns, taken through Ceph's group snapshot: nothing stops and the cluster never notices. Local-backend datastores have no block-level group primitive, so they fall outside the instant and are reported as such. The instant is crash-consistent, not application-consistent: if you need a specific write to be IN the snapshot, sync it first.

create

Snapshot every rbd-backed datastore in the cluster at ONE instant. Nothing is stopped. Any local-backend datastore in the cluster is listed as outside the instant: it is not part of the snapshot. The instant is crash-consistent, not application-consistent: if you need a specific write to be IN the snapshot, sync it first.

rdc cluster snapshot create <cluster> [options]

Flag	Description	Required	Default
`--snapshot <label>`	Snapshot label (default: a UTC timestamp)	No	-
`--debug`	Enable debug output	No	-

list

List the cluster's group snapshots.

rdc cluster snapshot list <cluster> [options]

Flag	Description	Required	Default
`--debug`	Enable debug output	No	-

5. MCP Server

Expose the CLI as a Model Context Protocol (MCP) server for integration with AI agents. The server wraps CLI commands as MCP tools with structured JSON input/output. Key subcommand: serve (start stdio-based MCP server).

5.1 serve

Start MCP server (stdio transport)

rdc mcp serve [options]

Flag	Description	Required	Default
`--config <name>`	Config name to use for all commands	No	-
`--timeout <ms>`	Default command timeout in milliseconds	No	`120000`

6. Repository Lifecycle

Deploy and manage encrypted repositories on machines. Key subcommands: create, up (deploy), down (stop), fork (instant CoW clone), delete, push/pull (backup), sync (file transfer). Lifecycle: repo create → repo up → repo down. See Key Concepts in rdc --help for architecture, routing, and environment variable details.

6.1 up

Deploy or update a repository (mount, run Rediaccfile up which calls renet compose). The machine is derived from the ref placement. Proxy routes take ~3s to become active after deploy. Prints the URL pattern for HTTP-exposed services (rediacc.service_port label) on completion. First deploy and forks are mounted automatically. CRIU checkpoint restore is auto-detected; use --skip-checkpoint to force fresh start. Use --all --machine <m> to deploy every repository on a machine.

rdc repo up [ref] [options]

Flag	Description	Required	Default
`--no-start`	Mount and prepare the repository without running its up() steps (folds the retired 'repo mount'; LUKS open and, for cluster repos, PV generation).	No	-
`--skip-checkpoint`	Skip CRIU checkpoint restore even if checkpoint data exists (force fresh start)	No	-
`--tls`	Request dedicated TLS cert for this repo (forks use shared machine cert by default)	No	-
`--no-wait`	Return once containers are started; health checks continue in the background	No	-
`--all`	Deploy every repository whose home is --machine (batch form)	No	-
`-m, --machine <name>`	With --all: the machine whose repositories to run the batch against	No	-
`--include-forks`	Also mount/start forked repositories	No	-
`--parallel`	Start repositories concurrently	No	-
`--concurrency <n>`	Max concurrent repositories (default: 3)	No	`3`
`-y, --yes`	Skip confirmation for batch operations	No	-
`--debug`	Enable debug output	No	-
`--skip-router-restart`	Skip restarting the route server after binary update	No	-
`--dry-run`	Show what would be done without making changes	No	-

6.2 down

Stop repository Docker containers (runs Rediaccfile down via renet compose). The machine is derived from the ref placement. Does NOT unmount the encrypted volume -- the repo stays mounted and can be restarted with 'repo up'. Use --unmount to also close the LUKS container after stopping (folds the retired 'repo unmount'). Use --checkpoint to save CRIU process state before stopping (next 'repo up' auto-restores). Use --all --machine <m> to stop every repository on a machine.

rdc repo down [ref] [options]

Flag	Description	Required	Default
`--unmount`	Also unmount (close the LUKS container) after stopping. Required before 'repo resize' or to fully secure the volume	No	-
`--checkpoint`	Create CRIU checkpoint before stopping (save process memory state for later restore via 'repo up')	No	-
`--all`	Stop every repository whose home is --machine (batch form)	No	-
`-m, --machine <name>`	With --all: the machine whose repositories to run the batch against	No	-
`--parallel`	Start repositories concurrently	No	-
`--concurrency <n>`	Max concurrent repositories (default: 3)	No	`3`
`-y, --yes`	Skip confirmation for batch operations	No	-
`--debug`	Enable debug output	No	-
`--skip-router-restart`	Skip restarting the route server after binary update	No	-
`--dry-run`	Show what would be done without making changes	No	-

6.3 status

Get repository status (mount state, Docker daemon running, container count, disk usage)

rdc repo status <ref> [options]

Flag	Description	Required	Default
`--debug`	Enable debug output	No	-
`--skip-router-restart`	Skip restarting the route server after binary update	No	-

6.4 cat

Read a bounded window of a file in a repository to stdout (diagnostics go to stderr).

rdc repo cat <ref> [options]

Flag	Description	Required	Default
`--remote-file <path>`	File path to read, relative to the repository mount root	Yes	-
`--max-bytes <n>`	Maximum bytes to read and print (default 1 MiB, hard ceiling 50 MiB)	No	-
`--offset <n>`	Byte offset to start reading from	No	-
`--head <lines>`	Print only the first N lines (cannot combine with byte range)	No	-
`--tail <lines>`	Print only the last N lines (cannot combine with byte range)	No	-
`--stat`	Print only size, type, and modification time; read no content	No	-
`--force-binary`	Allow reading binary (NUL-containing) content	No	-
`--debug`	Enable debug output	No	-
`--skip-router-restart`	Skip restarting the route server after binary update	No	-

6.5 diff

Git-style file-level diff between two copy-on-write forked repositories. Reports Added, Modified, Deleted, and Renamed files. Diffs the repository given by the positional <ref> (the target / new side) against its parent, resolved from local config, or against an explicit --base repository (the base / old side). Metadata-only and size-independent: it diffs the encrypted LUKS images at the block level without decrypting them, so a 1 GB repo and a 100 GB repo diff in the same milliseconds.

rdc repo diff <ref> [options]

Flag	Description	Required	Default
`--base <ref>`	Repository to diff against (the base / old side); defaults to the parent of <ref>	No	-
`--name-only`	Print only changed file paths, one per line (no status letters)	No	-
`--stat`	Show per-file change magnitude (byte and block deltas) and totals	No	-
`--content [path]`	Show a unified text diff for a single file (requires a file path)	No	-
`--fast`	Trust the block filter; skip content-hash confirmation (may over-report Modified)	No	-
`--debug`	Enable debug output	No	-
`--skip-router-restart`	Skip restarting the route server after binary update	No	-

6.6 list

List repositories on a machine

rdc repo list [options]

Flag	Description	Required	Default
`-m, --machine <name>`	Target machine name	No	-
`--datastore <name>`	List the repositories in this datastore (resolved to whichever machine currently holds it)	No	-
`--debug`	Enable debug output	No	-
`--skip-router-restart`	Skip restarting the route server after binary update	No	-

6.7 create

Create a new repository. State its home once: a machine or a datastore.

rdc repo create <name> [options]

Flag	Description	Required	Default
`-m, --machine <name>`	Target machine name	No	-
`--datastore <name>`	Named datastore that holds the repo (docker tiering, or the cluster form)	No	-
`--size <size>`	Repository size (e.g., 10G, 100G, 1T)	No	-
`--no-docker`	Skip starting Docker daemon after creation	No	-
`--debug`	Enable debug output	No	-
`--skip-router-restart`	Skip restarting the route server after binary update	No	-

6.8 delete

Delete a repository and its data. Config entry is preserved; use --archive-config to move credentials to deletedRepositories for recovery via 'repo admin archive restore'. Takes a positional <ref>; a bare name resolves to the grand line and is refused when several repos share the base name, so pass name:tag to target a fork.

rdc repo delete <ref> [options]

Flag	Description	Required	Default
`--archive-config`	Move config entry to deletedRepositories for later recovery	No	-
`-y, --yes`	Skip confirmation prompt	No	-
`--debug`	Enable debug output	No	-
`--skip-router-restart`	Skip restarting the route server after binary update	No	-
`--dry-run`	Show what would be done without making changes	No	-

6.9 commit

Freeze the current state of a mounted working fork into a new immutable commit (git-like). The commit records its message, author, timestamp, and parent in-volume (so it travels on push) and is marked read-only — it refuses to mount. The working fork continues unchanged, like git leaving the working tree intact. Check a commit out with 'rdc repo checkout' to get a writable copy.

rdc repo commit <ref> [options]

Flag	Description	Required	Default
`--message <msg>`	Commit message	Yes	-
`--author <author>`	Commit author	No	-
`--debug`	Enable debug output	No	-

6.10 branch

Create a named branch ref pointing at the working fork's current commit (its tip). Branch refs live in the CLI config (machine = object store, config = ref store).

rdc repo branch <ref> [options]

Flag	Description	Required	Default
`--branch <branch>`	Name of the new branch	Yes	-

6.11 checkout

Reflink-clone an immutable commit (or a branch tip) into a fresh writable working fork and point HEAD at it. Near-instant and constant-time (BTRFS reflink).

rdc repo checkout <commit-or-branch-ref> [options]

Flag	Description	Required	Default
`--tag <name>`	Name for the new writable working fork	Yes	-
`--from <workingFork>`	Resolve the positional <commit-or-branch-ref> as a branch name on this working fork	No	-
`--debug`	Enable debug output	No	-
`--skip-router-restart`	Skip restarting the route server after binary update	No	-

6.12 log

Print the commit history reachable from a working fork's current commit (or a commit reference), walking the parent chain recorded by 'rdc repo commit'. Reads the out-of-volume mirror, so no commit is unlocked.

rdc repo log <ref> [options]

Flag	Description	Required	Default
`--debug`	Enable debug output	No	-

6.13 merge

Merge a source commit or fork into a target working fork. The live target is never mutated in place: the result is built in a reflink clone and atomically swapped in. A mounted or running target is refused unless --force, which cleanly quiesces it first. Without --resolve it is a whole-image take-theirs (the target becomes the source); with --resolve ours|theirs it is a per-file three-way merge against the common ancestor, taking each side's unique changes and resolving two-sided conflicts per the flag.

rdc repo merge <ref> [options]

Flag	Description	Required	Default
`--from <source>`	Source commit or fork to merge from	Yes	-
`--force`	Quiesce a mounted/running target first, then merge (never mutates a live mount)	No	-
`—resolve <ours	theirs>`	Per-file conflict resolution for a three-way merge: 'ours' keeps the target's version, 'theirs' takes the source's. Omit for whole-image fast-forward (take-theirs).	No
`--base <guid>`	Common-ancestor commit GUID for a three-way merge (used with --resolve). Defaults to the source commit's parent or the target's current commit.	No	-
`--debug`	Enable debug output	No	-

6.14 gc

Delete immutable commit objects on a machine that no branch or HEAD reaches (reachability GC). The machine is the object store; the CLI config is the ref store. Dry-run by default: pass --apply to delete. Never touches a mounted object or a working fork.

rdc repo gc [options]

Flag	Description	Required	Default
`-m, --machine <name>`	Target machine name	Yes	-
`--apply`	Actually delete the unreachable commits (default is a dry-run preview)	No	-
`--debug`	Enable debug output	No	-

6.15 fork

Create a CoW (Copy-on-Write) fork of a repository. FORK IS NEAR-INSTANT AND CONSTANT-TIME regardless of repo size, BTRFS reflink clones the underlying image so a 100 GB repo and a 1 GB repo fork in the same ~seconds. The fork gets a NEW GUID, networkId, IP range, and auto-route domain ({service}-fork-{tag}.{repo}.{machine}.{baseDomain}) and is a fully independent copy. Online forking is supported, the parent can remain running. Fork inherits the parent's encryption credentials automatically. Use --checkpoint to capture CRIU process state before forking, the fork will auto-restore on first 'repo up' (in-memory state preserved). CROSS-MACHINE FORK: fork locally first, then transfer: (1) rdc repo fork <parent-ref> --tag <name>, (2) rdc repo push <fork-ref> --to <target-machine>, (3) rdc backup restore <fork-ref> --as <fork-name> -m <target-machine> --up. WARNING: do NOT use "repo push" alone to fork, it creates a raw copy with the SAME GUID (not an independent fork). Always fork first to get a new identity. Auto-routes use the repo name, so each fork gets a unique domain automatically.

rdc repo fork <ref> [options]

Flag	Description	Required	Default
`--tag <name>`	Tag for the fork (creates name:tag)	Yes	-
`--checkpoint`	Create CRIU checkpoint on source before forking (capture process memory state for restore on fork)	No	-
`--immutable`	Mark the fork read-only: it refuses to mount, keeping its image byte-stable forever (a frozen commit/base for cross-machine delta push)	No	-
`--up`	Mount and start services after forking (fork + mount + up in one command)	No	-
`--no-wait`	With --up: return once containers are started; health checks continue in the background (ideal for throwaway forks)	No	-
`--debug`	Enable debug output	No	-
`--skip-router-restart`	Skip restarting the route server after binary update	No	-

6.16 replicate

Create N read replicas of a repository in seconds, regardless of database size. Takes ONE datastore snapshot, then makes N constant-time fork-attaches (copy-on-write, throwaway local writes) spread across the cluster's nodes, and applies a generated overlay: one pinned PersistentVolume per replica, a StatefulSet, and two Services ({repo}-rw to the primary, {repo}-ro across the replicas). Replicas are POINT-IN-TIME copies: no replication stream, no read-your-writes; each replica runs one crash-recovery pass before serving. Refresh them with 'repo replicate refresh'. Replica sets are managed state: see 'repo replicate status' and remove with 'repo replicate remove'.

status

Show managed replica sets: replicas, hosting nodes, fork datastores, snapshot, and refresh timestamps.

rdc repo replicate status <ref>

CLI reference: rdc repo replicate status

remove

Remove a replica set: delete its generated k8s objects (label-scoped), discard its fork datastores, drop its snapshot, and forget the state. Infra steps are best-effort so removal converges even on a partially-gone cluster.

rdc repo replicate remove <ref> [options]

Flag	Description	Required	Default
`--debug`	Enable debug output	No	-

refresh

Roll every replica onto a fresh point-in-time snapshot, ONE at a time: bounce the replica pod, discard and re-fork its datastore under the same path, and let readiness re-admit it. N-1 replicas keep serving throughout.

rdc repo replicate refresh <ref> [options]

Flag	Description	Required	Default
`--debug`	Enable debug output	No	-

6.17 canary

Run a NEW image next to the stable one and split live traffic between them by percentage (release ladder rung 2). Creates one canary Deployment + Service on the SAME live data (no fork: canary users on forked data would read stale data and write into a doomed copy) and tells the Rediacc proxy to route the given percent of the stable hostname's traffic to it. Weight 0 keeps the canary dark; weight 100 is the blue/green flip. Before every canary change, an automatic group snapshot of the cluster's datastores is taken (release ladder rung 0) as the universal undo. Schema compatibility between the two versions (expand-contract) is the application's responsibility. For schema-BREAKING releases, use blue/green instead: fork the repo (instant, includes data), point the fork's Service here with weight 100, and roll back by restarting the untouched parent; writes made after the flip exist only in the new copy, so the rollback window is a policy decision.

status

Show managed canary sets: weight, images, stable service, and the latest release-undo snapshot.

rdc repo canary status <ref>

CLI reference: rdc repo canary status

weight

Change the percent of traffic routed to a canary. Takes a fresh release-undo group snapshot first, then re-applies the overlay; the proxy picks up the new split on its refresh tick. Weight 100 sends all traffic to the new version (the blue/green flip).

rdc repo canary weight <ref> [options]

Flag	Description	Required	Default
`--weight <percent>`	Percent of traffic routed to the canary (integer 0-100; 0 = dark, 100 = full flip)	Yes	-
`--debug`	Enable debug output	No	-

remove

Remove a canary's Deployment and Service (label-scoped); the stable Service serves 100% again. Release-undo snapshots are retained; prune them with the datastore snapshot commands.

rdc repo canary remove <ref> [options]

Flag	Description	Required	Default
`--debug`	Enable debug output	No	-

create

Run a NEW image next to the stable one and send it a percentage of live traffic. The canary shares the repository's live data (no fork), so schema compatibility between the two versions is the application's responsibility. A group snapshot of the cluster's datastores is taken first, as the undo.

rdc repo canary create <ref> [options]

Flag	Description	Required	Default
`--image <image>`	New image the canary Deployment runs (against shared live data)	Yes	-
`--port <port>`	Port the app serves on (must match the stable Service)	Yes	-
`--weight <percent>`	Percent of traffic routed to the canary (integer 0-100; 0 = dark, 100 = full flip)	Yes	-
`--service <name>`	Stable Service to split traffic with (default: the repo name)	No	-
`--replicas <n>`	Canary Deployment replica count (default: 1)	No	-
`--debug`	Enable debug output	No	-

6.18 resize

Resize a repository offline (supports both grow and shrink). Repo must be unmounted first ('repo down --unmount'). For zero-downtime growth without stopping, use 'repo expand' instead

rdc repo resize <ref> [options]

Flag	Description	Required	Default
`--size <size>`	New repository size (e.g., 10G, 100G, 1T)	Yes	-
`--debug`	Enable debug output	No	-
`--skip-router-restart`	Skip restarting the route server after binary update	No	-

6.19 expand

Expand a mounted repository online (zero downtime, grow-only). Grows the LUKS container and filesystem while containers keep running. Cannot shrink -- use 'repo resize' for that (requires unmount)

rdc repo expand <ref> [options]

Flag	Description	Required	Default
`--size <size>`	New repository size (e.g., 10G, 100G, 1T)	Yes	-
`--debug`	Enable debug output	No	-
`--skip-router-restart`	Skip restarting the route server after binary update	No	-

6.20 trim

Reclaim datastore pool space from mounted repositories (online, zero downtime). Frees blocks deleted inside repositories back to the pool via fstrim. Without a ref, trims every mounted repository on -m plus the datastore itself. fstrim is skipped for repositories under an active backup; --docker reclaim still runs

rdc repo trim [ref] [options]

Flag	Description	Required	Default
`--docker`	Reclaim Docker space (stopped containers, dangling images, build cache); runs even while a backup snapshot is active	No	-
`--docker-volumes`	Additionally prune unused Docker volumes	No	-
`--report-only`	Show discard state and reclaimable estimate without trimming	No	-
`-m, --machine <name>`	Target machine name	No	-
`--debug`	Enable debug output	No	-

6.21 policy

Manage the automatic size policy applied by the machine's storage maintainer: grow a repository's quota online when it fills up (never past max-quota, never when the pool is low) and run scheduled trims. There is no automatic shrink; use 'repo resize' for that. Without a ref, the machine-wide default policy on -m is addressed

set

Set size policy fields. Only the flags you pass are changed; other stored fields keep their values. Enabling auto-grow requires --max-quota: the ceiling is your explicit consent to over-provision the pool

rdc repo policy set [ref] [options]

Flag	Description	Required	Default
`--auto-grow <bool>`	Enable automatic online quota growth (true/false)	No	-
`--max-quota <size>`	Auto-grow ceiling (e.g. 200G); required for auto-grow	No	-
`--grow-threshold <percent>`	Filesystem used % that triggers a grow (default 85)	No	-
`--grow-step <step>`	Growth per step: absolute (10G) or percent of quota (20%)	No	-
`--auto-trim <bool>`	Enable scheduled trim (true/false)	No	-
`--trim-interval <hours>`	Minimum hours between automatic trims (default 24)	No	-
`-m, --machine <name>`	Target machine name	No	-
`--debug`	Enable debug output	No	-

get

Show the stored machine default, the repository override (with a ref), and the effective merged policy the maintainer acts on

rdc repo policy get [ref] [options]

Flag	Description	Required	Default
`-m, --machine <name>`	Target machine name	No	-
`--debug`	Enable debug output	No	-

6.22 push

Push repository to a remote (machine or storage). The target type is auto-detected from config. For machine-to-machine transfer, the encrypted repo image is copied with the SAME GUID — this is a backup/migration, not a fork. To create an independent fork, use 'repo fork' first, then push. A pushed copy lands as a backup ARTIFACT: boot it on the target with 'backup restore <ref> --as <name> -m <target> --up'

rdc repo push <ref> [options]

Flag	Description	Required	Default
`--to <remote>`	Destination machine or storage name (auto-detected from config)	No	-
`--to-machine <machine>`	—	No	-
`--provision <provider>`	Auto-provision target machine via cloud provider if it doesn't exist	No	-
`--checkpoint`	Create CRIU checkpoint before backup (captures process memory state for live migration)	No	-
`--force`	Force overwrite existing backup	No	-
`-w, --watch`	Watch for changes	No	-
`--bwlimit <limit>`	Bandwidth limit for rsync transfer (e.g., "6M", "10M")	No	-
`--delta-base <guid>`	Immutable base GUID present byte-identical on both machines; transfer only changed blocks (machine target). Omit for hands-free auto-base	No	-
`--strategy <strategy>`	Block-delta strategy when using a delta base: auto, physical, or shared	No	-
`--debug`	Enable debug output	No	-
`--skip-router-restart`	Skip restarting the route server after binary update	No	-

6.23 pull

Pull repository from a remote (machine or storage). Omit name to pull all repos. The source type is auto-detected from config. Use --up to deploy after pull

rdc repo pull <ref> [options]

Flag	Description	Required	Default
`--from <remote>`	Source machine or storage name (auto-detected from config)	No	-
`--from-machine <machine>`	—	No	-
`--force`	Force overwrite existing repository	No	-
`--up`	After pull, mount and deploy repository on this machine	No	-
`-w, --watch`	Watch for changes	No	-
`--bwlimit <limit>`	Bandwidth limit for rsync transfer (e.g., "6M", "10M")	No	-
`--delta-base <guid>`	Immutable base GUID present byte-identical on both machines; receive only changed blocks (machine source)	No	-
`--strategy <strategy>`	Block-delta strategy when using a delta base: auto, physical, or shared	No	-
`--debug`	Enable debug output	No	-
`--skip-router-restart`	Skip restarting the route server after binary update	No	-

6.24 migrate

Live-migrate a repository from one machine to another with minimal downtime. Two-phase rsync: bulk transfer while running, then brief stop for delta sync. Moves the whole repository (routing is repointed at the new home) and, once the move succeeds, deletes the source images (use --keep-source to retain them). Supports CRIU checkpoint for process memory migration and auto-provisioning of target machines

rdc repo migrate <ref> [options]

Flag	Description	Required	Default
`--to <place>`	Destination machine or cluster	Yes	-
`--provision <provider>`	Auto-provision target via cloud provider (e.g., hetzner, linode)	No	-
`--bwlimit <limit>`	Bandwidth limit for rsync transfer (e.g., 10M)	No	-
`--checkpoint`	CRIU live migration: capture and restore process memory state	No	-
`--delta-base <guid>`	Immutable base GUID for the cutover delta (advanced; defaults to the Phase-1 base)	No	-
`--strategy <strategy>`	Block-delta strategy for the cutover: auto, physical, or shared	No	-
`--skip-dns`	Skip DNS record switching after migration	No	-
`--keep-source`	Keep the source images after a successful move (by default they are deleted; leftover copies are not addressable by any config record and reconcile flags them as strays)	No	-
`--debug`	Enable debug output	No	-

6.25 sync

Sync repositories and files (bulk push/pull, file transfer)

upload

Upload files to a repository via rsync over SSH (delta transfer). Use --mirror to delete extra remote files, --dry-run to preview, --exclude to skip patterns

rdc repo sync upload <ref> [options]

Flag	Description	Required	Default
`--local <paths...>`	One or more local file or directory paths (default: current directory)	No	-
`--remote <path>`	Remote subdirectory path within repository	No	-
`--remote-file <path>`	Single remote file path (relative to repo root) to upload to. Mutually exclusive with --remote.	No	-
`--mirror`	Mirror mode - delete remote files not present locally	No	-
`--verify`	Verify files using checksums after sync	No	-
`--confirm`	Preview changes and ask for confirmation before syncing	No	-
`--exclude <patterns...>`	Patterns to exclude from sync	No	-
`--dry-run`	Show what would be done without making changes	No	-

download

Download files from a repository via rsync over SSH (delta transfer). Use --mirror to delete extra local files, --dry-run to preview, --exclude to skip patterns

rdc repo sync download <ref> [options]

Flag	Description	Required	Default
`--local <path>`	Local directory path (default: current directory)	No	-
`--remote <path>`	Remote subdirectory path within repository	No	-
`--remote-file <path>`	Remote file path within repository (alternative to --remote for single-file transfers)	No	-
`--mirror`	Mirror mode - delete local files not present on remote	No	-
`--verify`	Verify files using checksums after sync	No	-
`--confirm`	Preview changes and ask for confirmation before syncing	No	-
`--exclude <patterns...>`	Patterns to exclude from sync	No	-
`--dry-run`	Show what would be done without making changes	No	-

status

Dry-run comparison of local and remote files (shows what would be transferred without actually transferring)

rdc repo sync status <ref> [options]

Flag	Description	Required	Default
`--local <path>`	Local directory path (default: current directory)	No	-
`--remote <path>`	Remote subdirectory path within repository	No	-
`--remote-file <path>`	Remote file path within repository (alternative to --remote for single-file transfers)	No	-

6.26 tunnel

Create an SSH port-forward tunnel to a container's port on a remote machine. Auto-detects container and port when unambiguous. The tunnel stays open until you press Ctrl+C

rdc repo tunnel <ref> [options]

Flag	Description	Required	Default
`-c, --container <name>`	Container name (auto-detected if only one running)	No	-
`--port <port>`	Remote container port to forward	No	-
`--local <port>`	Local port (defaults to same as remote port)	No	-
`--url-only`	Print only the local URL once the tunnel is ready (machine-readable)	No	-

6.27 secret

Manage per-repo secrets injected at deploy time. Write-only: get returns digest only (never plaintext), set/unset require --current digest match or --rotate-secret. Two delivery modes: env (REDIACC_SECRET_<KEY>) and file (tmpfs at /var/run/rediacc/secrets/<networkId>/<KEY> via compose `secrets:`). Secrets never enter the encrypted LUKS image; forks inherit none.

get

Show the SHA-256 digest of a secret. The plaintext value is never returned by design (write-only). Use --current on a subsequent set/unset to verify a value you already know, or rotate via `set --rotate-secret`.

rdc repo secret get <ref> [options]

Flag	Description	Required	Default
`--key <KEY>`	Secret key in UPPER_SNAKE_CASE (max 64 chars). Will be exposed as REDIACC_SECRET_<KEY> for env-mode or /run/secrets/<key> in containers for file-mode.	Yes	-

list

List secret keys and modes (never values, never digests).

rdc repo secret list <ref>

CLI reference: rdc repo secret list

set

Set or overwrite a secret. Forks do not inherit; set on the fork explicitly. Under agent context, requires --current digest match (passwd-style).

rdc repo secret set <ref> [options]

Flag	Description	Required	Default
`--key <KEY>`	Secret key in UPPER_SNAKE_CASE (max 64 chars). Will be exposed as REDIACC_SECRET_<KEY> for env-mode or /run/secrets/<key> in containers for file-mode.	Yes	-
`--value <value>`	Secret value. Pass `-` to read from stdin (avoids shell-history exposure).	Yes	-
`--mode <mode>`	Delivery mode: 'env' (visible in container env, docker inspect) or 'file' (tmpfs file, never in env). Default: file.	No	`file`
`--current <value>`	Previous plaintext value (passwd-style precondition). Required for overwrite/unset; mutually exclusive with --rotate-secret.	No	-
`--rotate-secret`	Acknowledge rotation; skip --current precondition (audited as rotation). Use when intentionally rotating without verifying the prior value.	No	-

unset

Delete a secret. Under agent context, requires --current digest match.

rdc repo secret unset <ref> [options]

Flag	Description	Required	Default
`--key <KEY>`	Secret key in UPPER_SNAKE_CASE (max 64 chars). Will be exposed as REDIACC_SECRET_<KEY> for env-mode or /run/secrets/<key> in containers for file-mode.	Yes	-
`--current <value>`	Previous plaintext value (passwd-style precondition). Required for overwrite/unset; mutually exclusive with --rotate-secret.	No	-
`--rotate-secret`	Acknowledge rotation; skip --current precondition (audited as rotation). Use when intentionally rotating without verifying the prior value.	No	-

6.28 admin

Repository administration and maintenance verbs.

validate

Validate repository integrity (LUKS container, filesystem consistency, configuration). Use after unexpected shutdowns or to verify backup health

rdc repo admin validate <ref> [options]

Flag	Description	Required	Default
`--debug`	Enable debug output	No	-
`--skip-router-restart`	Skip restarting the route server after binary update	No	-

fsck

Validate the CLI config refs (branches, HEAD) against the objects actually present on a machine. Reports dangling refs (a ref pointing at a missing object) and orphan commits (an immutable commit no ref reaches). Read-only.

rdc repo admin fsck [options]

Flag	Description	Required	Default
`-m, --machine <name>`	Target machine name	Yes	-

ownership

Change repository directory ownership UID on the mounted volume (default: 7111). Use when containers need a specific UID to access repo files

rdc repo admin ownership <ref> [options]

Flag	Description	Required	Default
`--uid <uid>`	Owner UID (default: 7111)	No	-
`--debug`	Enable debug output	No	-
`--skip-router-restart`	Skip restarting the route server after binary update	No	-

autostart

Manage automatic repository startup on boot

enable:

Enable autostart for a repository (omit name to enable all)

rdc repo admin autostart enable [ref] [options]

Flag	Description	Required	Default
`--debug`	Enable debug output	No	-
`--skip-router-restart`	Skip restarting the route server after binary update	No	-

disable:

Disable autostart for a repository (omit name to disable all)

rdc repo admin autostart disable [ref] [options]

Flag	Description	Required	Default
`--debug`	Enable debug output	No	-
`--skip-router-restart`	Skip restarting the route server after binary update	No	-

list:

List repositories with autostart enabled

rdc repo admin autostart list [options]

Flag	Description	Required	Default
`-m, --machine <name>`	Target machine name	Yes	-
`--debug`	Enable debug output	No	-
`--skip-router-restart`	Skip restarting the route server after binary update	No	-

template

Manage and apply deployment templates (writes docker-compose.yml, Rediaccfile, and additional files). Rediaccfile must use 'renet compose' (not 'docker compose'). Restart policies are safe (auto-managed by renet watchdog). Renet auto-injects network_mode:host, CRIU capabilities, and rediacc labels. Dangerous settings (privileged, pid:host) are blocked. Use 'list' to see available built-in templates, 'apply' to deploy one to a repository

list:

List all embedded deployment templates shipped with the CLI

rdc repo admin template list

CLI reference: rdc repo admin template list

apply:

Apply a template to a repository. Use a built-in template name (e.g. app-postgres) or --file for a custom JSON template. Rediaccfile lifecycle: up() starts containers (pull images, generate configs here), down() stops. Minimal Rediaccfile: up() { renet compose -- pull; renet compose -- up -d; } down() { renet compose -- down; }. IMPORTANT: Rediaccfile MUST use 'renet compose': 'docker compose' is rejected. ENV VARS: two levels: (a) Rediaccfile shell: ${SVCNAME_IP} (e.g. APP_IP), ${REDIACC_WORKING_DIR}, ${REDIACC_NETWORK_ID}. (b) Inside containers: renet auto-injects SERVICE_IP and REDIACC_NETWORK_ID env vars. eBPF bind rewriting handles IP isolation transparently, so apps can bind to 0.0.0.0 and the kernel rewrites it to the correct loopback IP. Health checks can use localhost. network_mode:host is injected and ports: are ignored. STORAGE: Both ${REDIACC_WORKING_DIR}/... bind mounts and Docker named volumes are safe: Docker data-root is inside the encrypted LUKS mount. RESTART POLICY: Restart policies are safe: renet auto-strips them for CRIU compatibility and the watchdog handles recovery. Compose: do NOT add network_mode or rediacc.* labels (renet injects them). Multi-project: place each sub-project in its own subdirectory with its own Rediaccfile: renet auto-discovers and runs them in order. HTTPS routing: (A) Auto-route (fork-friendly, recommended): do NOT add traefik.enable. Renet auto-generates https://{serviceName}.{repoName}.{machineName}.{baseDomain}. Add rediacc.service_port=<port> label for non-80 ports. Each fork gets a unique domain. (B) Traefik labels (custom domain, NOT fork-friendly): traefik.enable=true, traefik.http.routers.<n>.rule=Host(`domain`), traefik.http.routers.<n>.entrypoints=websecure,websecure-v6, traefik.http.routers.<n>.tls.certresolver=letsencrypt, traefik.http.services.<n>.loadbalancer.server.port=<port>. For TCP/UDP: rediacc.tcp_ports=3306 / rediacc.udp_ports=53

rdc repo admin template apply <ref> [options]

Flag	Description	Required	Default
`--template <name>`	Template to apply	Yes	-
`--file <path>`	Path to custom template JSON file ({"version":"2","files":{"Rediaccfile":"...","docker-compose.yml":"..."}}): overrides the built-in template name	No	-
`--grand <name>`	Parent credential repository (auto-resolves name to GUID). Only for repos sharing secrets with a parent	No	-
`--debug`	Enable debug output	No	-
`--skip-router-restart`	Skip restarting the route server after binary update	No	-

6.29 promote

Make a validated fork the production repository under its parent name. The parent keeps its identity (GUID, networkId, domains, autostart, backup chain) and receives the fork's data; the old production data is preserved as a backup fork. Use it to test an upgrade on a fork, verify it, then promote. Pass an explicit <name>:<tag> for the fork; a bare ref resolves to the parent and is rejected with "not a fork". Promote never fetches bytes: use 'repo push' or 'backup restore' for that.

rdc repo promote <fork-ref> [options]

Flag	Description	Required	Default
`-y, --yes`	Skip confirmation prompt	No	-
`--debug`	Enable debug output	No	-
`--skip-router-restart`	Skip restarting the route server after binary update	No	-

6.30 logs

Show application logs from a repository's containers. Docker repositories read from the repository's own Docker daemon; kubernetes repositories read the pod logs in the repository's namespace. If the repository runs more than one container, name the one you want with --container.

rdc repo logs <ref> [options]

Flag	Description	Required	Default
`-c, --container <name>`	Container to read (only needed when the repository runs more than one)	No	-
`-f, --follow`	Keep streaming new lines until interrupted	No	-
`--lines <n>`	How many lines of history to show (default: 100)	No	`100`
`--timestamps`	Prefix each line with its timestamp	No	-
`--debug`	Enable debug output	No	-

6.31 exec

Run a command inside a repository container. The command's own exit code is passed straight through, so this works in scripts and conditionals. Docker repositories run it through the repository's Docker daemon; kubernetes repositories run it in the repository's namespace.

rdc repo exec <ref> <cmd...> [options]

CLI reference: rdc repo exec

Flag	Description	Required	Default
`-c, --container <name>`	Container to run in (only needed when the repository runs more than one)	No	-
`-i, --interactive`	Allocate a terminal (for a command that expects one)	No	-
`-u, --user <user>`	User to run the command as	No	-
`--debug`	Enable debug output	No	-

7. Detached Jobs

Manage detached jobs on a machine: list them, check status, replay or follow their event logs, cancel one, or garbage-collect finished ones. A detached job runs under systemd on the machine, so it survives a dropped connection and can be re-attached to later.

7.1 list

List detached jobs on a machine, newest first

rdc job list [options]

Flag	Description	Required	Default
`-m, --machine <name>`	Target machine name	Yes	-

7.2 status

Show a detached job's status

rdc job status <job-id> [options]

Flag	Description	Required	Default
`-m, --machine <name>`	Target machine name	Yes	-

7.3 logs

Replay or follow a detached job's event log

rdc job logs <job-id> [options]

Flag	Description	Required	Default
`-m, --machine <name>`	Target machine name	Yes	-
`-f, --follow`	Stream until the job finishes	No	-
`--since-line <n>`	Resume after this many lines (the count you already received)	No	-
`--debug`	Enable debug output	No	-

7.4 cancel

Cancel a running detached job

rdc job cancel <job-id> [options]

Flag	Description	Required	Default
`-m, --machine <name>`	Target machine name	Yes	-
`-y, --yes`	Skip confirmation prompt	No	-

7.5 gc

Remove finished jobs and their logs

rdc job gc [options]

Flag	Description	Required	Default
`-m, --machine <name>`	Target machine name	Yes	-
`--older-than <duration>`	Collect jobs that finished longer ago than this, in hours (default: 168)	No	-
`-y, --yes`	Skip confirmation prompt	No	-

8. Backup & Restore

Manage backup strategies, their scheduled runs, and the artifacts they produce: declare a strategy for a repository, run or restore it on demand, and inspect the archives it wrote to a storage endpoint.

8.1 schedule

Deploy backup schedule to a remote machine (systemd timers)

rdc backup schedule [options]

Flag	Description	Required	Default
`-m, --machine <name>`	Machine name	Yes	-
`--dry-run`	Preview generated units without deploying	No	-
`--force`	Proceed even if a backup is currently running (new unit applies on next tick; running invocation keeps its old unit)	No	-
`--reset-failed`	Clear failed state on touched services after a successful deploy (off by default, preserves failure signal)	No	-
`--debug`	Enable debug output	No	-

8.2 cancel

Cancel a running backup on a remote machine

rdc backup cancel [strategy] [options]

Flag	Description	Required	Default
`-m, --machine <name>`	Machine name	Yes	-
`--debug`	Enable debug output	No	-

8.3 list

List backup artifacts on a machine or storage.

rdc backup list [artifact-ref] [options]

Flag	Description	Required	Default
`-m, --machine <name>`	Machine name	No	-
`--storage <name>`	List artifacts on this storage endpoint	No	-
`--path <subdir>`	Subdirectory within the storage root. When omitted, hot/ and cold/ are listed and merged.	No	-
`-w, --watch`	Watch for changes	No	-
`--debug`	Enable debug output	No	-

8.4 restore

Turn a backup artifact into a live repository. Placement is stated here.

rdc backup restore <artifact-ref> [options]

Flag	Description	Required	Default
`--as <name>`	Name for the restored repository (defaults to the artifact name)	No	-
`-m, --machine <name>`	Machine name	No	-
`--datastore <name>`	Restore into this named datastore (its attached machine hosts it)	No	-
`--up`	Deploy the restored repository after the transfer	No	-
`--health-window <seconds>`	Total health-gate window in seconds (default 300)	No	-
`--health-timeout <seconds>`	Per-attempt health-check timeout in seconds (default 30)	No	-
`-y, --yes`	Skip confirmation prompt	No	-
`--debug`	Enable debug output	No	-

8.5 run

Run a backup now.

rdc backup run [strategy] [options]

Flag	Description	Required	Default
`-m, --machine <name>`	Machine name	Yes	-
`-w, --watch`	Watch for changes	No	-
`--debug`	Enable debug output	No	-

8.6 status

Show backup status and timer state on a remote machine

rdc backup status [strategy] [options]

Flag	Description	Required	Default
`-m, --machine <name>`	Machine name	Yes	-
`--debug`	Enable debug output	No	-

8.7 strategy

Manage named backup strategies (hot/cold modes, destinations, scheduling)

set

Create or update a backup strategy

rdc backup strategy set <strategy> [options]

Flag	Description	Required	Default
`--destination <name>`	Destination name within the strategy	No	-
`--storage <name>`	Storage config name (rclone credentials)	No	-
`--cron <expression>`	Cron schedule (e.g., "0 * * * *" for hourly)	No	-
`--mode <mode>`	Backup mode: "hot" (zero downtime) or "cold" (stop, snapshot, restart)	No	-
`--bwlimit <limit>`	Rclone bandwidth limit (e.g., "6M", "10M:off", "08:00,3M;22:00,10M")	No	-
`--include <repos>`	Only back up these repos (comma-separated names)	No	-
`--exclude <repos>`	Exclude these repos from backup (comma-separated names)	No	-
`--folder <path>`	Subfolder under the storage bucket for this destination (e.g. hot, cold)	No	-
`--enable`	Enable the strategy or destination	No	-
`--disable`	Disable the strategy or destination	No	-

bind

Bind a backup strategy to a machine so its schedule is deployed there

rdc backup strategy bind <strategy> [options]

Flag	Description	Required	Default
`-m, --machine <name>`	Target machine name	Yes	-

unbind

Unbind a backup strategy from a machine

rdc backup strategy unbind <strategy> [options]

Flag	Description	Required	Default
`-m, --machine <name>`	Target machine name	Yes	-

remove

Remove a backup strategy or destination

rdc backup strategy remove <strategy> [options]

Flag	Description	Required	Default
`--destination <name>`	Remove only this destination (keeps other destinations)	No	-

list

List all backup strategies

rdc backup strategy list

CLI reference: rdc backup strategy list

show

Show backup strategy details

rdc backup strategy show [strategy]

CLI reference: rdc backup strategy show

9. Storage Management

Manage external backup storage endpoints (S3 and other rclone-compatible targets): register and remove them, list what is registered, browse the artifacts they hold, and prune orphaned backups.

9.1 list

List storage endpoints. Give a name for full detail.

rdc storage list [name] [options]

Flag	Description	Required	Default
`--reveal`	Reveal the storage configuration in plaintext (interactive TTY only; audited)	No	-

9.2 browse

Browse files in a storage system

rdc storage browse <storage> [options]

Flag	Description	Required	Default
`--path <subpath>`	Subdirectory path to list	No	“

9.3 prune

Delete orphaned backups from storage that are no longer in any config. Multi-config safe with grace period protection. The rclone calls run on --machine (the executor), not on your laptop, so clients don't need rclone installed locally; --machine is the executor, not the source of truth.

rdc storage prune <storage> [options]

Flag	Description	Required	Default
`-m, --machine <name>`	Executor machine — runs the rclone list/delete calls against the storage. Required because clients aren't expected to have rclone installed locally; storage credentials still come from your local config.	Yes	-
`--dry-run`	Show what would be done without making changes	No	-
`--force`	Skip confirmation prompts	No	-
`--force-delete-mounted`	Override the mount-safety check and delete cloud backups even if the source GUID is currently mounted or has a running container on the executor machine. Distinct from --force (which only overrides the grace period for archived repos).	No	-
`--grace-days <days>`	Grace period in days for recently archived repos (default: 7)	No	-
`--debug`	Enable debug output	No	-
`--skip-router-restart`	Skip restarting the route server after binary update	No	-

9.4 add

rdc storage add <name> [options]

Flag	Description	Required	Default
`--vault <json>`	Vault content as JSON string	Yes	-

9.5 import

Import a storage endpoint from a definition file.

rdc storage import <file> [options]

Flag	Description	Required	Default
`--name <name>`	Import only this named section	No	-

9.6 remove

Remove a storage endpoint from the config.

rdc storage remove <name> [options]

Flag	Description	Required	Default
`-y, --yes`	Skip confirmation prompt	No	-
`--dry-run`	Show what would be done without making changes	No	-

10. VS Code Integration

Open VS Code with Remote SSH to a machine or repository with sandboxed per-repo environment. Same SSH connection as `term` but launches VS Code instead. Key subcommands: connect (or positional shorthand), list (show configured connections), cleanup (remove SSH config entries), check (verify VS Code installation).

10.1 connect

Open VS Code on a machine or inside a repository.

rdc vscode connect <target> [options]

Flag	Description	Required	Default
`-f, --folder <path>`	Remote folder path to open	No	-
`--url-only`	Print the VS Code URI instead of launching	No	-
`-n, --new-window`	Open in new VS Code window	No	-
`--skip-env-setup`	Skip remote environment setup	No	-
`--insiders`	Use VS Code Insiders settings	No	-
`--browser`	Serve browser VS Code from inside the repo sandbox (no local VS Code needed)	No	-
`--no-open`	Print the URL without launching the local browser	No	-
`--local <port>`	Local port (defaults to same as remote port)	No	-
`--server-provider <id>`	Browser VS Code server implementation (openvscode, code-server)	No	-
`--server-archive <file>`	Pre-staged server tarball path on the machine (airgapped installs)	No	-

10.2 list

List configured VS Code SSH connections

rdc vscode list

CLI reference: rdc vscode list

10.3 cleanup

Remove VS Code SSH configurations

rdc vscode cleanup [options]

Flag	Description	Required	Default
`--all`	Remove all rediacc SSH configurations	No	-
`-c, --connection <name>`	Remove specific connection	No	-

10.4 check

Check VS Code installation and configuration

rdc vscode check [options]

Flag	Description	Required	Default
`--insiders`	Use VS Code Insiders settings	No	-

10.5 serve

Manage the in-sandbox browser VS Code server

status

Show whether the browser VS Code server is running

rdc vscode serve status <target> [options]

Flag	Description	Required	Default
`--server-provider <id>`	Browser VS Code server implementation (openvscode, code-server)	No	-

stop

Stop the browser VS Code server

rdc vscode serve stop <target> [options]

Flag	Description	Required	Default
`--server-provider <id>`	Browser VS Code server implementation (openvscode, code-server)	No	-

11. SSH Terminal Access

Open an SSH terminal to a machine or a repository. The target is a machine name or a repo ref, and the repository context (DOCKER_HOST, working directory, environment) is set up automatically. Use `-c` to run a single command non-interactively. For container logs or a one-shot command in a container, use `repo logs` and `repo exec`. For VS Code, use `vscode`.

11.1 connect

Open a shell on a machine, or inside a repository with its Docker set up.

rdc term connect <target> [options]

Flag	Description	Required	Default
`-c, --command <cmd>`	Execute a command instead of interactive shell	No	-
`--external`	Force launching in external terminal window	No	-
`--reset-home`	Reset per-repo home overlay for a fresh start	No	-

12. Quick Commands (Shortcuts)

12.1 run

WARNING: Low-level command for debugging and experimentation only. Not for production use. Use higher-level commands (repo up, repo down, etc.) instead. Runs a raw bridge function on a machine.

rdc run

12.2 trace

Trace a task (shortcut for: queue trace)

rdc trace

12.3 cancel

Cancel a task (shortcut for: queue cancel)

rdc cancel

12.4 retry

Retry a failed task (shortcut for: queue retry)

rdc retry

13. Subscription Management

Manage licensing and subscription tokens. Sign in with login, inspect with status, reissue licenses with refresh.

Authenticate via browser or API token

rdc subscription login [options]

Flag	Description	Required	Default
`-t, --token <token>`	API token (rdt_...)	No	-
`--server <url>`	Account server URL	No	-

13.2 logout

Clear stored subscription token

rdc subscription logout

CLI reference: rdc subscription logout

13.3 status

Show subscription, and license state for a machine.

rdc subscription status [options]

Flag	Description	Required	Default
`-m, --machine <name>`	Machine name	No	-

13.4 refresh

Refresh licenses from the account server.

rdc subscription refresh [options]

Flag	Description	Required	Default
`-m, --machine <name>`	Machine name	No	-
`--repo <ref>`	Repository ref: name, or name:tag, optionally with @machine (for example shop or shop:test)	No	-

14. CLI Updates

Check for new CLI versions and apply updates. Supports --check-only to check without installing, and --force to update even if already on the latest version.

rdc update [options]

Flag	Description	Required	Default
`--force`	Force update even if already up-to-date	No	-
`--check-only`	Only check for updates without downloading	No	-
`--rollback`	Rollback to the previous version	No	-
`--status`	Show auto-update status and diagnostics	No	-
`--channel <channel>`	Set release channel (stable or edge)	No	-

15. Third-Party Credits

Print the third-party components bundled with rdc: the binaries renet embeds and conveys (CRIU, rsync, rclone), the embedded Node.js runtime, and the bundled npm dependencies. Shows versions, SPDX license identifiers, and the exact-version source archive URLs required for GPL compliance. Use --licenses to print the full THIRD_PARTY_LICENSES text bundled with release builds.

rdc credits [options]

Flag	Description	Required	Default
`--licenses`	Print the full THIRD_PARTY_LICENSES text bundled with release builds	No	-

16. System Diagnostics

Run diagnostic checks on the CLI environment: Node.js version, renet binary availability, config file validity, SSH key status, and network connectivity. Outputs a structured health report with pass/fail indicators.

rdc doctor

CLI reference: rdc doctor

17. Local VM Operations

Local VM provisioning and management using KVM/QEMU/Hyper-V (platform auto-detected). Not for remote machines — use `machine` for those. Key subcommands: up (start VMs), down (stop), status (list running VMs), ssh (connect to VM), setup (install prerequisites), check (verify host readiness).

17.1 up

Provision VM cluster locally

rdc ops up [options]

Flag	Description	Required	Default
`--force`	Force restart all VMs	No	-
`--parallel`	Create VMs in parallel	No	-
`--basic`	Minimal cluster (bridge + 1 worker)	No	-
`--lite`	Skip VM provisioning (status only)	No	-
`--skip-orchestration`	Skip cluster orchestration	No	-
`--backend <backend>`	Virtualization backend (kvm\|qemu, auto-detected)	No	-
`--os <name>`	VM operating system (e.g., ubuntu-24.04, debian-13)	No	-
`--debug`	Enable debug output	No	-

17.2 down

Destroy VM cluster

rdc ops down [options]

Flag	Description	Required	Default
`--backend <backend>`	Virtualization backend (kvm\|qemu, auto-detected)	No	-
`--debug`	Enable debug output	No	-

17.3 status

Show VM cluster status

rdc ops status [options]

Flag	Description	Required	Default
`--backend <backend>`	Virtualization backend (kvm\|qemu, auto-detected)	No	-

17.4 ssh

SSH into a VM

rdc ops ssh [options]

Flag	Description	Required	Default
`--vm-id <id>`	Virtual machine ID	Yes	-
`-c, --command <cmd>`	Execute a command instead of interactive shell	No	-
`--backend <backend>`	Virtualization backend (kvm\|qemu, auto-detected)	No	-
`--user <user>`	SSH username for VM connection	No	-

17.5 setup

Install virtualization prerequisites

rdc ops setup [options]

Flag	Description	Required	Default
`--debug`	Enable debug output	No	-

17.6 check

Verify virtualization prerequisites

rdc ops check

CLI reference: rdc ops check

18. Executor Service

Run this process as an executor so clients can run commands through it instead of connecting over SSH themselves

rdc serve [options]

Flag	Description	Required	Default
`-p, --port <port>`	Port to listen on	No	`8080`
`--host <host>`	Address to bind	No	`0.0.0.0`
`--mode <mode>`	Placement: daemon (customer host) or container	No	`daemon`

Common Error Messages

When something goes wrong, the CLI provides clear error messages:

Error	Meaning
Authentication required. Run: rdc subscription login	You need to log in first
No active config. Create one with: rdc config init <name>	No config is active — create one with `rdc config init`
Permission denied	Your account lacks the required permissions
Machine name required. Use --machine <name>	Specify a machine with `--machine` or set a config default
Team name required. Use --team or set default with: rdc config set team <name>	Specify a team with `--team` or set a config default
Region name required. Use --region or set default with: rdc config set region <name>	Specify a region with `--region` or set a config default

Output Formats

All list/get commands support multiple output formats via `--output`:

rdc machine list --output json
rdc machine list --output yaml
rdc machine list --output csv
rdc machine list --output table   # default

CLI reference: rdc machine list

This makes it easy to integrate Rediacc CLI into scripts, CI/CD pipelines, and automation workflows.