Initial commit of Node Affinity Anti-Affinity workload files #91
Conversation
|
@mffiedler @chaitanyaenr please review |
|
@mffiedler @chaitanyaenr please review, been running these workloads from Jenkins since OCP 4.2 |
|
|
||
| The Node Affinity Anti Affinity workload playbook is `workloads/node-affinity.yml` and will run the Node Affinity Anti Affinity Workload workload on your cluster. | ||
|
|
||
| The Node Affinity Anti Affinity workload's purpose is to validate if the OpenShift cluster can deploy 130 pause-pods with node affinity to one labeled worker node, and 130 hello-pods with anti-affinity to another labeled worker node. Deployed pods have memory and CPU requests, and goal is to be close 85% CPU capacity after all pods are deployed. |
There was a problem hiding this comment.
85% is subjective to the worker node instance type right?
| result_dir=${benchmark_results_dir} | ||
| fi | ||
|
|
||
| # git clone svt repo in /root |
There was a problem hiding this comment.
Maybe add a pointer to the script and pod spec/template in svt repo since they are being maintained there. This way, we will the location in case we need to modify something. On the other hand, it might be a good idea to bake in the script and pod templates into the workload itself like how we are doing in case of other workloads like nodevertical for example : https://github.com/openshift-scale/workloads/blob/master/workloads/templates/workload-nodevertical-script-cm.yml.j2#L75. This way we will have a clear idea of what workload is doing as well easier to modify it in case we want to change something. Thoughts?
Node Affinity Anti-Affinity workload files: