Add tests for validating recommendation json for accelerator values #1336
Conversation
| ("list_accelerator_recommendations", SUCCESS_STATUS_CODE, "v2.0", "human_eval_exp", "cluster-1", "resource-optimization-local-monitoring", "monitor", "local", "prometheus-1", "container", "statefulset", "human-eval-benchmark", "unpartitioned", None, None, "human-eval-benchmark", "15min", "0.1"), | ||
| ] | ||
| ) |
There was a problem hiding this comment.
Please share the test results and update the test documentation with the test requirements.
There was a problem hiding this comment.
This test is marked sanity so, it will run along with other kruize local tests on minikube and openshift. What is the behavior of this test when run on a non-GPU cluster?
There was a problem hiding this comment.
What is the behavior of this test when run on a non-GPU cluster?
This test should work fine for non-gpu usecase as I'm checking for presence of nvidia in recommendations. I haven't checked it for normal workload. WIll check it and update the test results here.
There was a problem hiding this comment.
@chandrams I have added the documentation as well. Can I please request your review? Thanks in advance!
| exp_name = input_json[0]['experiment_name'] | ||
|
|
||
| response = generate_recommendations(exp_name) |
There was a problem hiding this comment.
Shouldn't we run benchmark load here before generating the recommendations?
There was a problem hiding this comment.
This test is intended to run for a workload (typically a training or inference job) which is running for longer duration and has GPU usage.
|
As Github doesn't allow the upload of html files I have renamed it to |
|
@chandrams I have tested it on the Non-MIG GPU (Tesla T4) and the recommendations are as expected with no GPU recommendations and having only cpu and memory recommendations Uploading the report as the text file |
Signed-off-by: bharathappali <abharath@redhat.com>
Signed-off-by: bharathappali <abharath@redhat.com>
Signed-off-by: bharathappali <abharath@redhat.com>
bharathappali
force-pushed
the
gpu-support-pr-10
branch
from
4c413e8 to
5cb0c0c
Compare
|
|
||
| #### Prerequisites to run the test: | ||
|
|
||
| In addition to the pre-requisites mentioned above we need to make sure that a workload with name `human-eval-benchmark` is running in the namespace `unpartitioned` and has the accelerator usage data. |
There was a problem hiding this comment.
Please provide steps on how to run this workload or link to a doc that has the steps
There was a problem hiding this comment.
It's one of our benchmarks human-eval-benchmark can I add link to it?
There was a problem hiding this comment.
Added a link of human eval benchmark in the doc.
Signed-off-by: bharathappali <abharath@redhat.com>
Description
This PR adds tests for the accelerator values in list recommendation json
Type of change
How has this been tested?
Please describe the tests that were run to verify your changes and steps to reproduce. Please specify any test configuration required.
Test Configuration
Checklist 🎯
Additional information
Include any additional information such as links, test results, screenshots here