The AWS Deadline Cloud worker agent can be used to run a worker in an AWS Deadline Cloud fleet. This includes managing the life-cycle of a worker and its assigned work both in the service and on the worker's host.
Deadline Cloud schedules work as worker sessions which are an extension of Open Description (OpenJD) sessions specific to AWS Deadline Cloud. The worker agent initiates session actions, monitors them, and reports the status of running and completed session actions to the service including progress, logs, process exit code, and indicates if the work was canceled or interrupted.
The worker agent behavior follows the AWS Deadline Cloud worker API protocol that specifies the expectation of how the service and workers behave and collaborate through Deadline Cloud's worker APIs.
For guidance on setting up the worker agent for use in a customer-managed fleet, see the "Manage Deadline Cloud customer-managed fleets" topic in the AWS Deadline Cloud User Guide
The worker agent requires Python 3.9 or higher. There are additional platform-specific requirements listed below:
Linux:
- Amazon Linux 2 and 2023 are recommended and tested
sudo
must be installed
Windows:
- Windows Server 2022 is recommended and tested
- Requires CPython implementation of Python
- Python must be installed for all users (e.g. in
C:\Program Files
)
MacOS is intended to be used for testing only and is subject to change.
This package's version follows Semantic Versioning 2.0, but is still considered to be in its initial development, thus backwards incompatible versions are denoted by minor version bumps. To help illustrate how versions will increment during this initial development stage, they are described below:
- The MAJOR version is currently 0, indicating initial development
- The MINOR version is currently incremented when backwards incompatible changes are introduced to the public API.
- The PATCH version is currently incremented when bug fixes or backwards compatible changes are introduced to the public API.
We recommend installing the agent in a Python virtual environment (e.g. using venv
). For
this, run:
# Create venv
python -m venv /opt/deadline/worker
# Activate the virtual environment - you can later type "deactivate" to exit the environment
source /opt/deadline/worker/bin/activate
# Install worker agent program files into the virtual environment
pip install deadline-cloud-worker-agent
The worker agent runs as a Windows Service which leads to a few installation constraints:
- Python virtual environments are not supported
- Python must be installed for all users
To obtain the program files, run this command in an administrator command-prompt:
pip install deadline-cloud-worker-agent
The worker host can be prepared to be run using the provided install-deadline-worker
command. This
command performs certain functions to setup the worker host based on arguments provided. The
command performs all worker host setup activities, such as:
- creates an operating system user account (specified by the
--user
argument) on the worker host that the worker will run as.install-deadline-worker
accepts a previously created user. The user defaults todeadline-worker-agent
on Linux anddeadline-worker
on Windows. - creates a job user group (specified by
--group
, defaults todeadline-job-users
) if required. Theinstall-deadline-worker
accepts an existing group. - creates cache, log, and config directories, and an example config file
- [optionally] initializes the config file
- modifies the config file using provided arguments
- [optionally] install/update an operating system service
- [optionally] start the operating system service
NOTE: The install-deadline-worker
command does not support MacOS at this time.
To see the available command-line arguments, run:
On Linux:
Assuming you have installed the worker agent to a Python venv in /opt/deadline/worker
, run:
/opt/deadline/worker/bin/install-deadline-worker --help
On Windows:
Run the following command in an administrator command-prompt:
install-deadline-worker --help
Setting up the worker host using the install-deadline-worker
command (see "Setup worker host"
above) installs an operating system service. On Linux, this is a systemd service and on Windows this
is a Windows service.
The following commands demonstrate how to manually control the operating system service.
On Linux:
# Start the worker agent
systemctl start dealine-worker
# Stop the worker agent
systemctl stop dealine-worker
# Configure the worker agent to start on boot
systemctl enable dealine-worker
# Configure the worker agent to NOT start on boot
systemctl disable dealine-worker
On Windows:
Using an admin command-prompt:
REM start the service
sc.exe start DeadlineWorker
REM stop the service
sc.exe stop DeadlineWorker
See CONTRIBUTING.md
for information on reporting issues, requesting features, and developer information.
See security issue notifications for more information.
This project is licensed under the Apache-2.0 License.