Skip to content

'craneadm'

Usage:

$ 'craneadm' [OPTIONS] COMMAND [ARGS]...

Options:

  • -V, --version: Print Crane CLI version [default: False]
  • --install-completion: Install completion for the current shell.
  • --show-completion: Show completion for the current shell, to copy it or customize the installation.
  • --help: Show this message and exit.

Commands:

  • config: Configuration related commands.
  • demote: [Distributed] Demote a node to a follower...
  • init: [Distributed] Initialize crane cluster.
  • install: [Local] Pull Crane images of the given...
  • join: [Distributed] Join an existing cluster as a...
  • join-token: [Distributed] Print only the join token.
  • leave: [Distributed] Leave a cluster.
  • promote: [Distributed] Promote a node to a leader...
  • standalone: [Local] Launch a standalone cluster.
  • status: [Distributed] List all participating nodes in...

'craneadm' config

Configuration related commands.

Usage:

$ 'craneadm' config [OPTIONS] COMMAND [ARGS]...

Options:

  • --help: Show this message and exit.

Commands:

  • list: Get current crane url.
  • set: Set new configuration.

'craneadm' config list

Get current crane url.

Usage:

$ 'craneadm' config list [OPTIONS]

Options:

  • --help: Show this message and exit.

'craneadm' config set

Set new configuration.

Right now, only url can be configured.

Usage:

$ 'craneadm' config set [OPTIONS] VALUE

Arguments:

  • VALUE: [required]

Options:

  • --help: Show this message and exit.

'craneadm' demote

[Distributed] Demote a node to a follower node.

The node argument should be a unique prefix of the node's ID. If the node argument is not given, demotes self.

Usage:

$ 'craneadm' demote [OPTIONS]

Options:

  • --node TEXT
  • --help: Show this message and exit.

'craneadm' init

[Distributed] Initialize crane cluster.

Usage:

$ 'craneadm' init [OPTIONS]

Options:

  • --auth / --no-auth: Whether to enable authentication [default: True]
  • -g, --gpus TEXT: GPUs to use in this node, in CUDA_VISIBLE_DEVICES format, e.g. all / 0,2 / 2,4,8 [default: all]
  • --host-ip TEXT: IP address of this node [required]
  • --mount-local-crane TEXT: Setting this will mount your local crane directory to components
  • --rdma / --no-rdma: Whether to use RDMA [default: False]
  • --debug / --no-debug: Whether to enable debug mode [default: False]
  • --gateway-port INTEGER: The port that gateway listens to [default: 8000]
  • --ws-server-port INTEGER: The port that workspace server listens to [default: 8001]
  • --grafana-port INTEGER: The port that Grafana listens to [default: 3000]
  • --kibana-port INTEGER: The port that Kibana listens to [default: 5601]
  • --skip-monitor / --no-skip-monitor: Whether to skip monitoring stack. [default: False]
  • --skip-logging / --no-skip-logging: Whether to skip logging stack. [default: False]
  • --skip-billing / --no-skip-billing: Whether to skip billing stack. [default: False]
  • --skip-proxy / --no-skip-proxy: Whether to skip proxy server. [default: False]
  • --skip-workspace / --no-skip-workspace: Whether to skip workspace feature. [default: False]
  • --help: Show this message and exit.

'craneadm' install

[Local] Pull Crane images of the given version.

Version must be in semantic version format (e.g. 0.2.0), 'latest', or 'nightly'. 'latest' version refers to the latest stable release of crane.

TODO: master -> main

'nightly' version refers to the master branch.

Usage:

$ 'craneadm' install [OPTIONS] VERSION

Arguments:

  • VERSION: [required]

Options:

  • --help: Show this message and exit.

'craneadm' join

[Distributed] Join an existing cluster as a leader/follower.

Supply the IP address of the manager node and the docker swarm token. craneadm init would have given you a hint.

Usage:

$ 'craneadm' join [OPTIONS] SWARM_TOKEN MANAGER_IP

Arguments:

  • SWARM_TOKEN: [required]
  • MANAGER_IP: [required]

Options:

  • -g, --gpus TEXT: GPUs to use in this node, in CUDA_VISIBLE_DEVICES format, e.g. all / 0,2 / 2,4,8 [default: all]
  • --host-ip TEXT: IP address of this node [required]
  • --mount-local-crane TEXT: Setting this will mount your local crane directory to components
  • --rdma / --no-rdma: Whether to use RDMA [default: False]
  • --debug / --no-debug: Whether to enable debug mode [default: False]
  • --gateway-port INTEGER: The port that the gateway listens to [default: 8000]
  • --skip-monitor / --no-skip-monitor: Whether to skip monitoring stack. [default: False]
  • --skip-logging / --no-skip-logging: Whether to skip logging stack. [default: False]
  • --help: Show this message and exit.

'craneadm' join-token

[Distributed] Print only the join token.

Usage:

$ 'craneadm' join-token [OPTIONS]

Options:

  • --manager / --worker: Get worker or manager token. [default: False]
  • --help: Show this message and exit.

'craneadm' leave

[Distributed] Leave a cluster.

Usage:

$ 'craneadm' leave [OPTIONS]

Options:

  • --persist / --no-persist: Whether to delete container and volume [default: False]
  • --force / --no-force: Whether to force-leave cluster [default: False]
  • --help: Show this message and exit.

'craneadm' promote

[Distributed] Promote a node to a leader node.

The node argument should be a unique prefix of the node's ID.

Usage:

$ 'craneadm' promote [OPTIONS] NODE

Arguments:

  • NODE: [required]

Options:

  • --help: Show this message and exit.

'craneadm' standalone

[Local] Launch a standalone cluster.

Runs only single gateway process.

Usage:

$ 'craneadm' standalone [OPTIONS]

Options:

  • --auth / --no-auth: Whether to enable authentication [default: False]
  • -g, --gpus TEXT: Colon-separated CUDA_VISIBLE_DEVICES strings, e.g. 0,1:2,3 [default: ]
  • --state-backend [MEMORY|ETCD]: [default: MEMORY]
  • --port INTEGER: Gateway port. Random if not given.
  • --log-level TEXT: log level [default: info]
  • --help: Show this message and exit.

'craneadm' status

[Distributed] List all participating nodes in crane cluster.

Usage:

$ 'craneadm' status [OPTIONS]

Options:

  • --help: Show this message and exit.
Last update: March 2, 2022