With the release of botocore that includes our service model, our method of configuring service model is no longer sufficient to contact endpoints other than the one included in botocore. We need to set AWS_ENDPOINT_URL_DEADLINE to tell boto that we want requests to Deadline Cloud to go to a different endpoint.
Small issues:
The test fixtures code that copies the service model from the test runner machine to the worker machine does not support gzip files, which is the format that service models included botocore are in.
The KEEP_WORKER_AFTER_FAILURE configuration was not working, workers are still shutting down after failing to start
The WORKER_REGION configuration is unnecessarily specific to the worker. The region can be applied to all components created by the test fixtures.
What was the solution? (How)
Set AWS_ENDPOINT_URL_DEADLINE whenever we install a service model
Support gzip (.json.gz) service model files
The ServiceModel class was responsible for too many things. Removed the file paths and the AWS CLI commands from the class to simplify the code.
Make KEEP_WORKER_AFTER_FAILURE actually keep the worker around after failing
Deprecate WORKER_REGION and use REGION instead
What is the impact of this change?
We can contact different Deadline Cloud endpoints for our tests
What was the problem/requirement? (What/Why)
Main issue:
AWS_ENDPOINT_URL_DEADLINE
to tell boto that we want requests to Deadline Cloud to go to a different endpoint.Small issues:
KEEP_WORKER_AFTER_FAILURE
configuration was not working, workers are still shutting down after failing to startWORKER_REGION
configuration is unnecessarily specific to the worker. The region can be applied to all components created by the test fixtures.What was the solution? (How)
AWS_ENDPOINT_URL_DEADLINE
whenever we install a service model.json.gz
) service model filesServiceModel
class was responsible for too many things. Removed the file paths and the AWS CLI commands from the class to simplify the code.KEEP_WORKER_AFTER_FAILURE
actually keep the worker around after failingWORKER_REGION
and useREGION
insteadWhat is the impact of this change?
KEEP_WORKER_AFTER_FAILURE
worksWORKER_REGION
configuration is deprecatedHow was this change tested?
hatch build && hatch run test
Was this change documented?
No
Is this a breaking change?
Yes
By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.