Elasticseach performance testing for Origin Aggregate Logging

This project tests the performance of Elasticsearch deployed as a part of Origin Aggregate Logging component. It automatically synchronizes and deploys pods containing Rally. The pods are meant to run user specified tests against the Elasticsearch. Results of those tests can then be viewed in Grafana dashboard, which is also deployed by this tool.

Before using

Before using this project make sure you meet these requirements:

• Have Docker and Python 3 installed.
• Install the Python dependencies

$ pip3 install -r requirements.txt

• Have an access to an Openshift cluster and be logged in. Verify by executing:

$ oc get nodes
# The output should be a list of nodes.

• The Openshift cluster has to have a properly deployed Origin Aggregate Logging component. To verify this, go to Openshift web UI and access the Prometheus web UI. Then, navigate to Status and Targets tab. Find the Elasticsearch entry, based on the name of the service monitor in the openshift-logging project.
• Have an access to a container image repository where the script can upload an image.
• Have a folder containing all the necessary files for the Rally. You can use the demo folder start/esrally-container/copy/multiple.
• Start the script from the root directory of this project. Verify by executing:

$ ls start/
bootstrap-container  esrally-container

Quick Start

$ python3 py-scripts/perfStackDeploy.py REPOSITORY FOLDER

Example:

$ python3 py-scripts/perfStackDeploy.py docker.io/vladmasarik/hackfest-dep start/esrally-container/copy/multiple

For more help see:

python3 py-scripts/perfStackDeploy.py --help

Folder Architecture

• polling.go is used to build the synchronization binary placed into the Rally container.
• manifests contains configuration of all the objects that are used.
• py-scripts contains mainly perfStackDeploy.py which is the main script executing the systems logic, and label-for-bulk.py which is a script used to add metadata to the testing data.
• start/esrally-container contains Dockerfile used to create the testing container image that should be executed and copy has files that should be copied into the image.

Notes

• As of 20th of May the CLO does not deploy the ES ServiceMonitor properly. AFAIK the Prometheus does not see it in a openshift-logging project, and I was not able to fix it. You can still see the CPU and metrics usage of the ES though.
• The script does not delete the instances it deploys. Therefore, if you want to run the script again you need to delete those objects. oc delete -f manifests/ is more than enough. Secrets get replaced and so does the tarred testing file, therefore, you do not have to delete those.
• If you want to create your own testing track you should start with the Rally documentation.

License

Apache License 2.0