After you have installed the worker agent, installed the software necessary to process your jobs, and configured the AWS credentials for the worker agent, you should test that the installation can process your jobs before creating an AMI for your fleet. You should test the following:
-
The Deadline Cloud worker agent is properly configured to run as a system service.
-
That the worker polls the associated queue for work.
-
That the worker successfully processes jobs sent to the queue associated with the fleet.
After you test the configuration and can successfully process representative jobs, you can use the configured worker to create an AMI for Amazon EC2 workers, or as a model for your on-premise workers.
Note
If you are testing the worker host configuration of an auto scaling fleet, you may have difficulty testing your worker in the following situations:
-
If there is no work in the queue, Deadline Cloud stops the worker agent shortly after the worker starts.
-
If the worker agent is configured to shut down the host when stopping, the agent shuts down the machine if there is no work in the queue.
To avoid these issues, use a staging fleet that doesn't auto scale to configure and test your workers. After testing the worker host, be sure to set the correct fleet ID before baking an AMI.
To test your worker host configuration
-
Run the worker agent by starting the operating system service.
From a root shell run the following command:
systemctl start deadline-worker
-
Monitor the worker to make sure it starts and polls for work.
From a root shell run the following command:
systemctl status deadline-worker
The command should return a response like:
Active: active (running) since Wed 2023-06-14 14:44:27 UTC; 7min ago
If the response doesn't look like that, inspect the log file using the following command:
tail -n 25 /var/log/amazon/deadline/worker-agent.log
-
Submit jobs to queue associated with your fleet. The jobs should be representative of the jobs that the fleet processes.
-
Monitor the progress of the job using the Deadline Cloud monitor or CLI. If a job fails, check the session and worker logs.
-
Update the configuration of the worker host as needed until jobs complete successfully.
-
When the test jobs succeed you can stop the worker:
From a root shell run the following command:
systemctl stop deadline-worker