-
Epic
-
Resolution: Unresolved
-
Undefined
-
None
-
None
-
None
-
None
-
Collect Perf & Scale results from baseline cluster (AWS IPI)
-
False
-
-
False
-
In Progress
-
80% To Do, 20% In Progress, 0% Done
- As an OpenShift engineer, I would like to gather automatically the performance data from standard performance profiles from kube-burner on CI workflow (baseline clusters), so that I know what is expected performance data from the well-know/controlled environment to provide a broad overview/guidance to partner's operation when validating OpenShift clusters in their infrastructure.
DESCRIPTION
The Perf & Scale team created a guide[1] defining the detailed steps to run benchmark tests in the Control and Data Plane. The guide clearly defines the input (commands to run) and expected results (SLO).
To achieve the goal of helping the partners to define the "baseline 'flavors' for production" in a given cloud provider, we want to collect the standard (light profiles) from CI executions so we will have an overview of what to expect as baseline performance data to of OpenShift installations in controlled/well-known environments/infra (OCP CI) beforehand exposing it to partners.
Eventually we can expose/collect that performance data, but not block the review pipeline without guarantee of well-defined SLO rules on OPCT.
ACCEPTANCE CRITERIA
- Collect kube-burner profiles in the collector plugins
- Expose an option to opt-out the kube-burner in the collector plugin (while the feature does not have blocker SLO)
OPEN QUESTIONS
- Should we need to opt-out by default to only CLI consumes it? it will require more work in the CLI to create a flag to enable it in the 'run' command.
REFERENCES
- is related to
-
OPCT-244 Collect etcd performance
-
- To Do
-