oslat.md

oslat

oslat is a test program for detecting OS level thread latency caused by unexpected system scheduling or interruptions (e.g., system ticks). The goal of the oslat workload in the benchmark-operator is to run oslat inside of a container and measure the latency per core. This is

Running oslat

Given that you followed instructions to deploy operator, you can modify cr.yaml to your needs. It is recommended to define pod requests and limits when running oslat test, to give guaranteed CPUs to the pods. It is also expected to have the realtime kernel installed with required isolation for pods using the Performance Add-On Operator.

The option runtime_class can be set to specify an optional runtime_class to the podSpec runtimeClassName. This is primarily intended for Kata containers.

The option annotations can be set to apply the specified annotations to the pod metadata.

An example CR might look like this

apiVersion: ripsaw.cloudbulldozer.io/v1alpha1
kind: Benchmark
metadata:
  name: oslat
  namespace: benchmark-operator
spec:
  elasticsearch:
    server: <ES_SERVER>
  workload:
    name: "oslat"
    args:
      node_selector: "<nodeSelector for the RT worker>"
      runtime: "1m"
      disable_cpu_balance: true
      use_taskset: true
      pod:
        requests:
          memory: "200Mi"
          cpu: "4"
        limits:
          memory: "200Mi"
          cpu: "4"

You can run it by:

# kubectl apply -f config/samples/oslat/cr.yaml # if edited the original one
# kubectl apply -f <path_to_file> # if created a new cr file

Looking at results

You can look at the results from the oslat benchmark by doing

NAME=oslat-workload-xxxxxxx
oc logs -n benchmark-operator $NAME

Cleanup

When you're done simply delete the benchmark CR; the job and its pod will be garbage-collected automatically.