This PR enhances the deploy_utils.py file to support asynchronous execution of the pipeline_run method using run_in_threadpool. This change allows the server to handle multiple pipeline requests concurrently, improving performance and efficiency.
I've already tested it locally with a docker-compose.yml with
where my_custom_deploy_utils.py is the adjusted file (with run_in_threadpool etc)
If you need additional testing resources, I've created a dummy component that just waits, some pipeline and send_requests_and_save.py that makes multiple async request in order
previously
This PR enhances the deploy_utils.py file to support asynchronous execution of the
pipeline_run
method usingrun_in_threadpool
. This change allows the server to handle multiple pipeline requests concurrently, improving performance and efficiency.I've already tested it locally with a
docker-compose.yml
withwhere
my_custom_deploy_utils.py
is the adjusted file (with run_in_threadpool etc)If you need additional testing resources, I've created a dummy component that just waits, some pipeline and
send_requests_and_save.py
that makes multiple async request in order previouslyReturned syncronous result of the format
(pipeline 4 starts after all tasks from pipeline 1 are ended)
With the changes, it returned me
which look intertwined (async)
test_pipelines.zip