aws-deadline/deadline-cloud-worker-agent/worker-agent-testing

Deadline Cloud Worker Agent Testing

This skill provides guidance for running tests in the deadline-cloud-worker-agent repository.

Unit Tests

Run unit tests with:

hatch run test

Run tests across all supported Python versions:

hatch run all:test

Linting & Formatting

hatch run lint
hatch run fmt

Integration Tests

Integration tests run locally on the host machine and cover OS/filesystem integration. If changes apply to both Windows and Linux, test on both platforms.

hatch run integ-test

E2E Tests

E2E tests run the agent on EC2 instances against the live Deadline Cloud service. Tests are under test/e2e/.

Prerequisites

1. Configure AWS credentials and region:

``sh export AWS_PROFILE=<your-profile> export AWS_DEFAULT_REGION=<region> ``

1. Deploy the S3 DeleteBucketContents CloudFormation resource to your account:

https://github.com/aws-cloudformation/community-registry-extensions/blob/main/resources/S3_DeleteBucketContents/resource-role-prod.yaml Note the output role ARN.

1. Activate AwsCommunity::S3::DeleteBucketContents in AWS Console → CloudFormation → Public Extensions (use the role ARN from step 2).
2. Ensure your account has sufficient Farm quota (limit: 2 per account).
3. Deploy testing infrastructure:

``sh scripts/deploy_e2e_testing_infrastructure.sh ``

1. Gather environment variable exports per OS:

``sh ./scripts/get_e2e_test_ids_from_cfn.sh --os Linux > .e2e_linux_infra.sh ./scripts/get_e2e_test_ids_from_cfn.sh --os Windows > .e2e_windows_infra.sh ``

Running E2E Tests

rm -f dist/*
hatch build
export WORKER_AGENT_WHL_PATH=$(pwd)/$(ls dist/*.whl)
 
# Linux
source .e2e_linux_infra.sh
hatch run e2e:test
 
# Windows
source .e2e_windows_infra.sh
hatch run e2e:test

Using Local Wheel Files for Dependencies

You can override the openjd-sessions and/or deadline-cloud packages installed on the worker by pointing to local wheel files. This is useful when testing against unreleased or locally-built versions of those packages.

Both variables accept a path (glob patterns are supported, but must resolve to exactly one file). The wheel is copied to the worker's temp directory and installed via pip alongside the worker agent.

# Override openjd-sessions with a local build
export OPENJD_SESSIONS_WHL_PATH=/path/to/openjd_sessions-*.whl
 
# Override deadline-cloud with a local build
export DEADLINE_WHL_PATH=/path/to/deadline_cloud-*.whl

These can be combined with WORKER_AGENT_WHL_PATH:

export WORKER_AGENT_WHL_PATH=$(pwd)/$(ls dist/*.whl)
export OPENJD_SESSIONS_WHL_PATH=/path/to/openjd_sessions-*.whl
export DEADLINE_WHL_PATH=/path/to/deadline_cloud-*.whl
hatch run e2e:test

Debugging Pytest Hanging

If pytest hangs after tests complete (likely due to a non-exiting Python thread), enable thread stack debugging:

export DEBUG_THREAD_STACKS=1
hatch run e2e:test

Live Service Testing (Manual)

Required AWS Resources

• A Farm, Fleet, and Queue
• A Bootstrapping Role (needs CreateWorker and GetWorkerIamCredentials permissions; trust policy allows your account to assume it)
• A Worker/Fleet Role (trust policy allows assume-role by the service's credential-vending service principal)
• Optionally, a Queue Role

Use the helper script to create non-role resources:

scripts/create_service_resources.sh <worker role arn> <queue role arn>

Running the Agent via Docker

1. Install the service model: aws configure add-model --service-name deadline ...

> Important: Use --service-name deadline (not any other name). The container mounts ~/.aws as a > volume and symlinks it so boto3 can find the model at its default search path. This is required when > testing against internal or beta endpoints whose URLs differ from the public service. > If testing against the public Deadline Cloud service, skip this step.

1. Set env vars:

``sh export FARM_ID=<farm-id> export FLEET_ID=<fleet-id> export AWS_DEFAULT_REGION=<region> # Bootstrapping role credentials: source scripts/assume_role_to_env.sh <role arn> ``

1. Run the agent:

``sh scripts/run_posix_docker.sh --build ``

1. Stop the agent:

``sh docker exec test_worker_agent /home/agentuser/term_agent.sh ``