x2i sends raw data from Gatling, JMeter and K6 load generation tools to InfluxDB time series database. x2i is the default solution for Perfana customers to send load generation tool results to Perfana and use this data to automatically analyse performance tests. x2i can also be used independently of Perfana.
x2i evolved from Anton Kramarev's gatling-to-influxdb, a project that processes the Gatling log file and sends the raw data to influxdb for analysis. Anton's goal was to replace the Graphite backend plugin in Gatling that allows aggregated data to send to influxdb. Using raw data allows for more detailed analysis of test results using Influxdb.
At Perfana we forked the project, added support for additional load generation tools, and made several other improvements. Hence the name 'x2i', where the 'x' stands for multiple load generation tools.
Existing Perfana customers and trial users best contact Perfana directly in case of any questions or issues. If you are not (yet ;-) a Perfana customer or trial user, feel free to create an issue in this Github repository. Please limit an issue to x2i specifically, we won't be able to troubleshoot your custom integration.
Binaries for multiple operating systems and architectures are provided. In production x2i is mostly used on AMD64 Linux. Our development systems also include MacOS on ARM. Other platforms are used sporadically and could have unknown issues.
x2i requires an existing database with read/write access. The read access is used to verify the connection to the database.
The following versions of InfluxDB are known to work:
Later versions of InfluxDB are not supported.
Run x2i -h for a quick help with examples for the different load generation tools that are supported.
x2i needs only one argument, the location where to find the results of the testrun. In a common scenario you will provide additional arguments for:
The --system-under-test and --test-environment arguments could be used, and are required for the Perfana integration
to work properly.
Datapoints are send to Influxdb when either the hardcoded timeout of 5 seconds is reached, or when --max-batch-size
number of datapoints is logged. The number of datapoints that is being send is logged in the x2i log file.
Add a View Results Tree component that logs errors and successes to the Filename
~/results/${__time(yyyy-MM-dd-HH-mm)}/resultsTree.csv. Configure it by enabling the following fields:
Use k6 run --out csv=test_results.csv.
When using x2i with Perfana, x2i will be started using a CommandRunnerEventConfig in your POM. In the starter
packages for gatling, jmeter and k6, you will find a perfect example. The --system-under-test and --test-environment
arguments are required for the Perfana integration to work properly.
You could integrate x2i in your CI/CD pipeline. When running in detached mode, x2i will return the PID as follows:
[PID] 20201allowing you to stop the process when the test has finished.
echo "Starting x2i in detached mode, saving PID in variable" && \
x2iPID=$(x2i <arguments> -d | awk '{print $2}') && \
echo "Build and execute test" && \
<run your test> \
echo "Waiting for parser to safely finish all its work" && \
sleep 30 && \
echo "Sending interrupt signal to x2i process" && \
kill -INT $x2iPID && \
echo "Waiting for process to stop safely" && \
sleep 30 && \
echo "Exiting"Building the application should be straightforward. Install the go language development tools and run go build.
We welcome contributions from users of x2i. If you would like to contribute to this project, we encourage you to create an issue to discuss the contribution first.
This application is licensed under MIT license. Please read the LICENSE file for details.