This repo is designed to provide a randomized load for "chaos testing".
Prioritization & project work are tracked in Work items
The generator can be run from the command-line or via a Helm chart.
Command line invocation:
source setup-env.sh
KUBECONFIG=/path/to/your/kubeconfig workload_runner.py
A number of options are available:
$ ./workload_runner.py --help
usage: workload_runner.py [-h] [-l LOG_DIR] [-m {ReadWriteOnce,ReadWriteMany}]
[-n NAMESPACE] [--oc OC] [--ocs-namespace OCS_NAMESPACE]
[-s STORAGECLASS] [-z]
optional arguments:
-h, --help show this help message and exit
-l LOG_DIR, --log-dir LOG_DIR
Path to use for log files
-m {ReadWriteOnce,ReadWriteMany}, --accessmode {ReadWriteOnce,ReadWriteMany}
StorageClassName for the workload's PVCs
-n NAMESPACE, --namespace NAMESPACE
Namespace to use for the workload
--oc OC Path/executable for the oc command
--ocs-namespace OCS_NAMESPACE
Namespace where the OCS components are running
-s STORAGECLASS, --storageclass STORAGECLASS
StorageClassName for the workload's PVCs
-z, --sleep-on-error On error, sleep forever instead of exit
Deployment via the Helm chart (using Helm v3):
$ KUBECONFIG=~/src/osio4/kubeconfig helm install foo helm/ocs-monkey-generator
NAME: foo
LAST DEPLOYED: 2019-07-25 15:31:07.04315199 -0400 EDT m=+0.615644812
NAMESPACE: default
STATUS: deployed
NOTES:
ocs-monkey workload generator has been deployed!
Generator is running in namespace: default
By default, the generator will create & destroy Deployments in the namespace
ocs-monkey
, stopping and collecting logs when it detects a problem.