An implementation of JMeter instances for remote testing on Kubernetes, forked and based on kubernauts/jmeter-kubernetes
The following features of Google Cloud Platform must be enabled first, namely: Google Compute Engine,Google Kubernetes Engine and Cloud Load Balancing
The recommended settings are:
- Machine Configuration not lower than E2 series
e2-small (2 vCPU, 2GB memory)per node. - Boot Disk Size not higher than
50 GB - Node Pool Size of
3 + <Number of Slaves>
This implementation requires at the very least 5 nodes,
- JMeter Master
- JMeter Slaves (Min. 2)
- InfluxDB
- Grafana
Therefore, if it is desired to have 4 slaves, the number of the required nodes are 7 nodes.
Once the cluster is created and verified as healthy, run the following command to get credentials to the cluster
gcloud container clusters get-credentials <your cluster name> --project=<your_project> --region=<your_cluster_region>
Then verify that it is selected via the following command
kubectl config get-contexts
If it is not selected, run the following command to switch context
kubectl config set current-context <your_jmeter_context>
Note that this will only work on UNIX or UNIX-like system, as the script requires bash as a shell in running these scripts.
Windows User could circumvent this problem by running the scripts within the Google's Cloud Shell and the Cloud Shell Editor
Clone this repository:
https://github.com/razyoboy/jmeter-kubernetes.git
cd jmeter-kubernetes
To change the number of slaves, modify the file jmeter_slave_deploy.yaml under the key specs/replicas
spec:
replicas: <number_of_slaves>
Note that the number of slaves must abide the number of nodes as mentioned above (No. of Nodes = 3 + <Number of Slaves> )
Run the creation script:
./jmeter_cluster_create.sh
Note however, that the script would store the namespace string within the file tenant_export within the repo's root, for further uses
Once created, verify that the workload are all up and running by using:
kubectl get po -n <your_namespace>
and verify that all are ready.
This will link the JMeter Master pod with InfluxDB and Grafana for Load Test Data Visualization
./dashboard.sh
Once created, the expected results would be along the lines of this:
Creating Influxdb jmeter Database
Creating the Influxdb data source
{"datasource":{"id":1,"orgId":1,"name":"jmeterdb","type":"influxdb","typeLogoUrl":"","access":"proxy","url":"http://jmeter-influxdb:8086","password":"admin","user":"admin","database":"jmeter","basicAuth":false,"basicAuthUser":"","basicAuthPassword":"","withCredentials":false,"isDefault":true,"secureJsonFields":{},"version":1,"readOnly":false},"id":1,"message":"Datasource added","name":"jmeterdb"}
The creation of dashboard component is now completed.
The script has been modified to use the Grafana Pod as a Load-Balancer, allowing access directly from a front-end IP address.
Once the Master and Slaves Creation and the Dashboard Creation is completed, you can now access Grafana and configure a few things
- Configure the DB source, select the InfluxDB's source as
jmeterdb - Configure or Import your JMeter Dashboard, alternatively you can copy & paste the following template GrafanaJMeterTemplate.json
Once all the above workloads are successfully created, a test plan can now be executed to the JMeter Master Pod, which can be run by using:
./start_test.sh <path_to_your_.jmx_test_plan>
The script has been modified from the oringinal to also include logging and exporting flags to the JMeter Master Pod, this allows the generation of HTML report, which shall be explained further below.
Once the test has been completed, the resulting <test_plan>.csv and <test_plan> folder will be created within the root directory of this repository (jmeter-kubernetes/Output)
To (force) stop the test, invoke Keyboard Interrupt (CTRL+C) and run the following script to send the termination request to all of the slaves connected to the master.
./jmeter_stop.sh