Running in AWS

Toil jobs can be run on a variety of cloud platforms. Of these, Amazon Web Services (AWS) is currently the best-supported solution. Toil provides the Cluster Utilities to conveniently create AWS clusters, connect to the leader of the cluster, and then launch a workflow. The leader handles distributing the jobs over the worker nodes and autoscaling to optimize costs.

The fastest way to get started with Toil in a cloud environment is by using Toil’s autoscaling capabilities to handle node provisioning. Autoscaling is a powerful and efficient tool for running your cluster in the cloud. It manages your cluster for you and scales up or down depending on the workflow’s demands.

The Running a Workflow with Autoscaling section details how to create a cluster and run a workflow that will dynamically scale depending on the workflow’s needs.

The Static Provisioning section explains how a static cluster (one that won’t automatically change in size) can be created and provisioned (grown, shrunk, destroyed, etc.).

To setup AWS, see Preparing your AWS environment.

Toil Provisioner

The Toil provisioner is included in Toil alongside the [aws] extra and allows us to spin up a cluster.

Getting started with the provisioner is simple:

  1. Make sure you have Toil installed with the AWS extras. For detailed instructions see Installing Toil with extra features.
  2. You will need an AWS account and you will need to save your AWS credentials on your local machine. For help setting up an AWS account see here. For setting up your aws credentials follow instructions here.

The Toil provisioner is built around the Toil Appliance, a Docker image that bundles Toil and all its requirements (e.g. Mesos). This makes deployment simple across platforms, and you can even simulate a cluster locally (see Developing with the Toil Appliance for details).

Choosing Toil Appliance Image

When using the Toil provisioner, the appliance image will be automatically chosen based on the pip installed version of Toil on your system. That choice can be overriden by setting the environment variables TOIL_DOCKER_REGISTRY and TOIL_DOCKER_NAME or TOIL_APPLIANCE_SELF. See Toil Environment Variables for more information on these variables. If you are developing with autoscaling and want to test and build your own appliance have a look at Developing with the Toil Appliance.

For information on using the Toil Provisioner have a look at Running a Workflow with Autoscaling.

Details about Launching a Cluster in AWS

Using the provisioner to launch a Toil leader instance is simple using the launch-cluster command. For example, to launch a cluster named “my-cluster” with a t2.medium leader in the us-west-2a zone, run:

(venv) $ toil launch-cluster my-cluster \
--leaderNodeType t2.medium \
--zone us-west-2a \
--keyPairName <your-AWS-key-pair-name>

The cluster name is used to uniquely identify your cluster and will be used to populate the instance’s Name tag. In addition, the Toil provisioner will automatically tag your cluster with an Owner tag that corresponds to your keypair name to facilitate cost tracking.

The leaderNodeType is an EC2 instance type. This only affects the leader node.

The --zone parameter specifies which EC2 availability zone to launch the cluster in. Alternatively, you can specify this option via the TOIL_AWS_ZONE environment variable. Note: the zone is different from an EC2 region. A region corresponds to a geographical area like us-west-2 (Oregon), and availability zones are partitions of this area like us-west-2a.

For more information on options try:

(venv) $ toil launch-cluster --help

Static Provisioning

Toil can be used to manage a cluster in the cloud by using the Cluster Utilities. The cluster utilities also make it easy to run a toil workflow directly on this cluster. We call this static provisioning because the size of the cluster does not change. This is in contrast with Running a Workflow with Autoscaling.

To launch worker nodes alongside the leader we use the -w option.:

(venv) $ toil launch-cluster my-cluster --leaderNodeType t2.small \
-z us-west-2a --keyPairName your-AWS-key-pair-name --nodeTypes m3.large,t2.micro -w 1,4

This will spin up a leader node of type t2.small with five additional workers - one m3.large instance and four t2.micro.

Now we can follow the instructions under Running in AWS to start the workflow on the cluster.

Currently static provisioning is only possible during the cluster’s creation. The ability to add new nodes and remove existing nodes via the native provisioner is in development, but can also be achieved through CGCloud. Of course the cluster can always be deleted with the destroy-cluster utility.

Note

CGCloud also can do static provisioning for an AWS cluster, however it is being phased out in favor of the Toil provisioner.

Uploading Workflows

Now that our cluster is launched, we use the rsync-cluster utility to copy the workflow to the leader. For a simple workflow in a single file this might look like:

(venv) $ toil rsync-cluster -z us-west-2a my-cluster toil-workflow.py :/

Note

If your toil workflow has dependencies have a look at the Deploying a Remote Workflow section for a detailed explanation on how to include them.

Running a Workflow with Autoscaling

Autoscaling is a feature of running Toil in a cloud whereby additional cloud instances are launched to run the workflow. Autoscaling leverages Mesos containers to provide an execution environment for these workflows.

  1. Download sort.py.

  2. Launch the leader node in AWS using the launch-cluster command.

    (venv) $ toil launch-cluster <cluster-name> \
    --keyPairName <AWS-key-pair-name> \
    --leaderNodeType t2.micro \
    --zone us-west-2a
    
  3. Copy the sort.py script up to the leader node.

    (venv) $ toil rsync-cluster <cluster-name> sort.py :/tmp
    
  4. Login to the leader node.

    (venv) $ toil ssh-cluster <cluster-name>
    
  5. Run the script as an autoscaling workflow.

    $ python /tmp/sort.py  \
    aws:us-west-2:autoscaling-sort-jobstore \
    --provisioner aws --nodeTypes c3.large --maxNodes 2\
    --batchSystem mesos --mesosMaster <private-IP>:5050
    --logLevel DEBUG
    

    Note

    In this example, the autoscaling Toil code creates up to two instances of type c3.large and launches Mesos slave containers inside them. The containers are then available to run jobs defined by the sort.py script. Toil also creates a bucket in S3 called aws:us-west-2:autoscaling-sort-jobstore to store intermediate job results. The Toil autoscaler can also provision multiple different node types, which is useful for workflows that have jobs with varying resource requirements. For example, one could execute the script with --nodeTypes c3.large,r3.xlarge --maxNodes 5,1, which would allow the provisioner to create up to five c3.large nodes and one r3.xlarge node for memory-intensive jobs. In this situation, the autoscaler would avoid creating the more expensive r3.xlarge node until needed, running most jobs on the c3.large nodes.

  6. View the generated file to sort.

    $ head fileToSort.txt
    
  7. View the sorted file.

    $ head sortedFile.txt
    

For more information on other autoscaling (and other) options have a look at Toil Workflow Options and Command Line Interface and/or run:

$ python my-toil-script.py --help

Important

Some important caveats about starting a toil run through an ssh session are explained in the ssh-cluster section.

Preemptability

Toil can run on a heterogeneous cluster of both preemptable and non-preemptable nodes. A node type can be specified as preemptable by adding a spot bid to its entry in the list of node types provided with the --nodeTypes flag. While individual jobs can each explicitly specify whether or not they should be run on preemptable nodes via the boolean preemptable resource requirement, the --defaultPreemptable flag will allow jobs without a preemptable requirement to run on preemptable machines.

Specify Preemptability Carefully

Ensure that your choices for --nodeTypes and --maxNodes <> make sense for your workflow and won’t cause it to hang. You should make sure the provisioner is able to create nodes large enough to run the largest job in the workflow, and that non-preemptable node types are allowed if there are non-preemptable jobs in the workflow.

Finally, the --preemptableCompensation flag can be used to handle cases where preemptable nodes may not be available but are required for your workflow. With this flag enabled, the autoscaler will attempt to compensate for a shortage of preemptable nodes of a certain type by creating non-preemptable nodes of that type, if non-preemptable nodes of that type were specified in --nodeTypes.

Using Mesos with Toil on AWS

The mesos master and agent processes bind to the private IP addresses of their EC2 instance, so be sure to use the master’s private IP when specifying --mesosMaster. Using the public IP will prevent the nodes from properly discovering each other.

Dashboard

Toil provides a dashboard for viewing the RAM and CPU usage of each node, the number of issued jobs of each type, the number of failed jobs, and the size of the jobs queue. To launch this dashboard for a toil workflow, include the --metrics flag in the toil script command. The dashboard can then be viewed in your browser at localhost:3000 while connected to the leader node through toil ssh-cluster. On AWS, the dashboard keeps track of every node in the cluster to monitor CPU and RAM usage, but it can also be used while running a workflow on a single machine. The dashboard uses Grafana as the front end for displaying real-time plots, and Prometheus for tracking metrics exported by toil. In order to use the dashboard for a non-released toil version, you will have to build the containers locally with make docker, since the prometheus, grafana, and mtail containers used in the dashboard are tied to a specific toil version.

Cluster Utilities

There are several utilities used for starting and managing a Toil cluster using the AWS provisioner. They are installed via the [aws] extra. For installation details see Toil Provisioner. The cluster utilities are used for Running in AWS and are comprised of toil launch-cluster, toil rsync-cluster, toil ssh-cluster, and toil destroy-cluster entry points. For a detailed explanation of the cluster utilities run:

toil --help

For information on a specific utility run:

toil launch-cluster --help

for a full list of its options and functionality.

Note

Boto must be configured with AWS credentials before using cluster utilities.

launch-cluster

Running toil launch-cluster starts up a leader for a cluster. Workers can be added to the initial cluster by specifying the -w option. For an example usage see launch-cluster. More information can be found using the --help option.

ssh-cluster

Toil provides the ability to ssh into the leader of the cluster. This can be done as follows:

$ toil ssh-cluster CLUSTER-NAME-HERE

This will open a shell on the Toil leader and is used to start an Running a Workflow with Autoscaling run. Issues with docker prevent using screen and tmux when sshing the cluster (The shell doesn’t know that it is a TTY which prevents it from allocating a new screen session). This can be worked around via:

$ script
$ screen

Simply running screen within script will get things working properly again.

Finally, you can execute remote commands with the following syntax:

$ toil ssh-cluster CLUSTER-NAME-HERE remoteCommand

It is not advised that you run your Toil workflow using remote execution like this unless a tool like nohup is used to insure the process does not die if the SSH connection is interrupted.

For an example usage, see Running a Workflow with Autoscaling.

rsync-cluster

The most frequent use case for the rsync-cluster utility is deploying your Toil script to the Toil leader. Note that the syntax is the same as traditional rsync with the exception of the hostname before the colon. This is not needed in toil rsync-cluster since the hostname is automatically determined by Toil.

Here is an example of its usage:

$ toil rsync-cluster CLUSTER-NAME-HERE \
   ~/localFile :/remoteDestination

destroy-cluster

The destroy-cluster command is the advised way to get rid of any Toil cluster launched using the launch-cluster command. It ensures that all attached node, volumes, and security groups etc. are deleted. If a node or cluster in shut down using Amazon’s online portal residual resources may still be in use in the background. To delete a cluster run

$ toil destroy-cluster CLUSTER-NAME-HERE