Tune Distributed Experiments

Tune is commonly used for large-scale distributed hyperparameter optimization. This page will overview how to setup and launch a distributed experiment along with commonly used commands for Tune when running distributed experiments.


To run a distributed experiment with Tune, you need to:

  1. First, start a Ray cluster if you have not already.

  2. Specify ray.init(address=...) in your script to connect to the existing Ray cluster.

  3. Run the script on the head node (or use ray submit).

Local Cluster Setup

If you already have a list of nodes, you can follow the local private cluster setup. Below is an example cluster configuration as tune-default.yaml:

cluster_name: local-default
    type: local
auth: {ssh_user: YOUR_USERNAME, ssh_private_key: ~/.ssh/id_rsa}
## Typically for local clusters, min_workers == max_workers.
min_workers: 3
max_workers: 3
setup_commands:  # Set up each node.
    - pip install ray torch torchvision tabulate tensorboard

ray up starts Ray on the cluster of nodes.

ray up tune-default.yaml

ray submit uploads tune_script.py to the cluster and runs python tune_script.py [args].

ray submit tune-default.yaml tune_script.py -- --ray-address=localhost:6379

Manual Local Cluster Setup

If you run into issues using the local cluster setup (or want to add nodes manually), you can use the manual cluster setup. At a glance,

Launching a cloud cluster


If you have already have a list of nodes, go to Local Cluster Setup.

Ray currently supports AWS and GCP. Follow the instructions below to launch nodes on AWS (using the Deep Learning AMI). See the cluster setup documentation. Save the below cluster configuration (tune-default.yaml):

cluster_name: tune-default
provider: {type: aws, region: us-west-2}
auth: {ssh_user: ubuntu}
min_workers: 3
max_workers: 3
# Deep Learning AMI (Ubuntu) Version 21.0
head_node: {InstanceType: c5.xlarge, ImageId: ami-0b294f219d14e6a82}
worker_nodes: {InstanceType: c5.xlarge, ImageId: ami-0b294f219d14e6a82}
setup_commands: # Set up each node.
    - pip install ray torch torchvision tabulate tensorboard

ray up starts Ray on the cluster of nodes.

ray up tune-default.yaml

ray submit --start starts a cluster as specified by the given cluster configuration YAML file, uploads tune_script.py to the cluster, and runs python tune_script.py [args].

ray submit tune-default.yaml tune_script.py --start -- --ray-address=localhost:6379

Analyze your results on TensorBoard by starting TensorBoard on the remote head machine.

# Go to http://localhost:6006 to access TensorBoard.
ray exec tune-default.yaml 'tensorboard --logdir=~/ray_results/ --port 6006' --port-forward 6006

Note that you can customize the directory of results by running: tune.run(local_dir=..). You can then point TensorBoard to that directory to visualize results. You can also use awless for easy cluster management on AWS.

Running a distributed experiment

Running a distributed (multi-node) experiment requires Ray to be started already. You can do this on local machines or on the cloud.

Across your machines, Tune will automatically detect the number of GPUs and CPUs without you needing to manage CUDA_VISIBLE_DEVICES.

To execute a distributed experiment, call ray.init(address=XXX) before tune.run, where XXX is the Ray redis address, which defaults to localhost:6379. The Tune python script should be executed only on the head node of the Ray cluster.

One common approach to modifying an existing Tune experiment to go distributed is to set an argparse variable so that toggling between distributed and single-node is seamless.

import ray
import argparse

parser = argparse.ArgumentParser()
args = parser.parse_args()

# On the head node, connect to an existing ray cluster
$ python tune_script.py --ray-address=localhost:XXXX

If you used a cluster configuration (starting a cluster with ray up or ray submit --start), use:

ray submit tune-default.yaml tune_script.py -- --ray-address=localhost:6379


  1. In the examples, the Ray redis address commonly used is localhost:6379.

  2. If the Ray cluster is already started, you should not need to run anything on the worker nodes.


Tune automatically syncs the trial folder on remote nodes back to the head node. This requires the ray cluster to be started with the cluster launcher. By default, local syncing requires rsync to be installed. You can customize the sync command with the sync_to_driver argument in tune.SyncConfig by providing either a function or a string.

If a string is provided, then it must include replacement fields {source} and {target}, like rsync -savz -e "ssh -i ssh_key.pem" {source} {target}. Alternatively, a function can be provided with the following signature:

def custom_sync_func(source, target):
    sync_cmd = "rsync {source} {target}".format(
    sync_process = subprocess.Popen(sync_cmd, shell=True)


When syncing results back to the driver, the source would be a path similar to ubuntu@, and the target would be a local path. This custom sync command is used to restart trials under failure. The sync_to_driver is invoked to push a checkpoint to new node for a paused/pre-empted trial to resume.

Pre-emptible Instances (Cloud)

Running on spot instances (or pre-emptible instances) can reduce the cost of your experiment. You can enable spot instances in AWS via the following configuration modification:

# Provider-specific config for worker nodes, e.g. instance type.
    InstanceType: m5.large
    ImageId: ami-0b294f219d14e6a82 # Deep Learning AMI (Ubuntu) Version 21.0

    # Run workers on spot by default. Comment this out to use on-demand.
        MarketType: spot
            MaxPrice: 1.0  # Max Hourly Price

In GCP, you can use the following configuration modification:

    machineType: n1-standard-2
      - boot: true
        autoDelete: true
        type: PERSISTENT
          diskSizeGb: 50
          # See https://cloud.google.com/compute/docs/images for more images
          sourceImage: projects/deeplearning-platform-release/global/images/family/tf-1-13-cpu

    # Run workers on preemtible instances.
      - preemptible: true

Spot instances may be removed suddenly while trials are still running. Often times this may be difficult to deal with when using other distributed hyperparameter optimization frameworks. Tune allows users to mitigate the effects of this by preserving the progress of your model training through checkpointing.

search_space = {
    "lr": tune.sample_from(lambda spec: 10**(-10 * np.random.rand())),
    "momentum": tune.uniform(0.1, 0.9)

analysis = tune.run(
    TrainMNIST, config=search_space, stop={"training_iteration": 10})

Example for using spot instances (AWS)

Here is an example for running Tune on spot instances. This assumes your AWS credentials have already been setup (aws configure):

  1. Download a full example Tune experiment script here. This includes a Trainable with checkpointing: mnist_pytorch_trainable.py. To run this example, you will need to install the following:

$ pip install ray torch torchvision filelock
  1. Download an example cluster yaml here: tune-default.yaml

  2. Run ray submit as below to run Tune across them. Append [--start] if the cluster is not up yet. Append [--stop] to automatically shutdown your nodes after running.

ray submit tune-default.yaml mnist_pytorch_trainable.py --start -- --ray-address=localhost:6379
  1. Optionally for testing on AWS or GCP, you can use the following to kill a random worker node after all the worker nodes are up

$ ray kill-random-node tune-default.yaml --hard

To summarize, here are the commands to run:

wget https://raw.githubusercontent.com/ray-project/ray/master/python/ray/tune/examples/mnist_pytorch_trainable.py
wget https://raw.githubusercontent.com/ray-project/ray/master/python/ray/tune/tune-default.yaml
ray submit tune-default.yaml mnist_pytorch_trainable.py --start -- --ray-address=localhost:6379

# wait a while until after all nodes have started
ray kill-random-node tune-default.yaml --hard

You should see Tune eventually continue the trials on a different worker node. See the Fault Tolerance section for more details.

You can also specify tune.run(sync_config=tune.SyncConfig(upload_dir=...)) to sync results with a cloud storage like S3, allowing you to persist results in case you want to start and stop your cluster automatically.

Fault Tolerance

Tune will automatically restart trials in case of trial failures/error (if max_failures != 0), both in the single node and distributed setting.

Tune will restore trials from the latest checkpoint, where available. In the distributed setting, if using the cluster launcher with rsync enabled, Tune will automatically sync the trial folder with the driver. For example, if a node is lost while a trial (specifically, the corresponding Trainable actor of the trial) is still executing on that node and a checkpoint of the trial exists, Tune will wait until available resources are available to begin executing the trial again.

If the trial/actor is placed on a different node, Tune will automatically push the previous checkpoint file to that node and restore the remote trial actor state, allowing the trial to resume from the latest checkpoint even after failure.

Recovering From Failures

Tune automatically persists the progress of your entire experiment (a tune.run session), so if an experiment crashes or is otherwise cancelled, it can be resumed by passing one of True, False, “LOCAL”, “REMOTE”, or “PROMPT” to tune.run(resume=...). Note that this only works if trial checkpoints are detected, whether it be by manual or periodic checkpointing.


  • The default setting of resume=False creates a new experiment.

  • resume="LOCAL" and resume=True restore the experiment from local_dir/[experiment_name].

  • resume="REMOTE" syncs the upload dir down to the local dir and then restores the experiment from local_dir/experiment_name.

  • resume="PROMPT" will cause Tune to prompt you for whether you want to resume. You can always force a new experiment to be created by changing the experiment name.

Note that trials will be restored to their last checkpoint. If trial checkpointing is not enabled, unfinished trials will be restarted from scratch.



Upon a second run, this will restore the entire experiment state from ~/path/to/results/my_experiment_name. Importantly, any changes to the experiment specification upon resume will be ignored. For example, if the previous experiment has reached its termination, then resuming it with a new stop criterion will not run. The new experiment will terminate immediately after initialization. If you want to change the configuration, such as training more iterations, you can do so restore the checkpoint by setting restore=<path-to-checkpoint> - note that this only works for a single trial.


This feature is still experimental, so any provided Trial Scheduler or Search Algorithm will not be checkpointed and able to resume. Only FIFOScheduler and BasicVariantGenerator will be supported.

Common Commands

Below are some commonly used commands for submitting experiments. Please see the Autoscaler page to see find more comprehensive documentation of commands.

# Upload `tune_experiment.py` from your local machine onto the cluster. Then,
# run `python tune_experiment.py --address=localhost:6379` on the remote machine.
$ ray submit CLUSTER.YAML tune_experiment.py -- --address=localhost:6379

# Start a cluster and run an experiment in a detached tmux session,
# and shut down the cluster as soon as the experiment completes.
# In `tune_experiment.py`, set `tune.SyncConfig(upload_dir="s3://...")`
# and pass it to `tune.run(sync_config=...)` to persist results
$ ray submit CLUSTER.YAML --tmux --start --stop tune_experiment.py -- --address=localhost:6379

# To start or update your cluster:
$ ray up CLUSTER.YAML [-y]

# Shut-down all instances of your cluster:
$ ray down CLUSTER.YAML [-y]

# Run Tensorboard and forward the port to your own machine.
$ ray exec CLUSTER.YAML 'tensorboard --logdir ~/ray_results/ --port 6006' --port-forward 6006

# Run Jupyter Lab and forward the port to your own machine.
$ ray exec CLUSTER.YAML 'jupyter lab --port 6006' --port-forward 6006

# Get a summary of all the experiments and trials that have executed so far.
$ ray exec CLUSTER.YAML 'tune ls ~/ray_results'

# Upload and sync file_mounts up to the cluster with this command.
$ ray rsync-up CLUSTER.YAML

# Download the results directory from your cluster head node to your local machine on ``~/cluster_results``.
$ ray rsync-down CLUSTER.YAML '~/ray_results' ~/cluster_results

# Launching multiple clusters using the same configuration.
$ ray up CLUSTER.YAML -n="cluster1"
$ ray up CLUSTER.YAML -n="cluster2"
$ ray up CLUSTER.YAML -n="cluster3"


Sometimes, your program may freeze. Run this to restart the Ray cluster without running any of the installation commands.

$ ray up CLUSTER.YAML --restart-only