| commit | 04cdd9f9dcfd20b7e6d8d475c0028aee2b623f68 | [log] [tgz] |
|---|---|---|
| author | Joey Armstrong <joey@opennetworking.org> | Fri Jun 09 15:18:23 2023 -0400 |
| committer | Joey Armstrong <joey@opennetworking.org> | Wed Jul 19 15:00:36 2023 -0400 |
| tree | 352dc7ba6ff3b9bb90771319bfdf0ea7b5721b42 | |
| parent | 9684bea2ea28ca5984c43a305c60380a4f5c622d [diff] |
[VOL-5032] - Build and deploy voltha-system-tests for v2.12 VERSION ======= o Bump version string to initiate a triage build for release. makefiles/ ========== o https://github.com/opencord/onf-make o Update to the latest version of onf-make library makefiles to enable additional linting targets. Change-Id: I4cdc959d616174573014bd6343825492b01e456b
Automated test-suites to validate the stability/functionality of VOLTHA. Tests that reside in here should be written in Robot Framework and Python.
Intended use includes:
Learn more about VOLTHA System Test in Test Automation Brigade.
Python 3.5 or later and virtualenv
voltctl - a command line tool to access VOLTHA. Reference - voltctl
kubectl - a command line tool to access your Kubernetes Clusters. Reference - kubectl
voltctl and kubectl must be properly configured on your system prior to any test executions.
Directory is structured as follows:
├── tests └── functional/ // feature/functionality tests, recovery tests and more, that are specific to ATT workflow └── dt-workflow/ // feature/functionality tests, recovery tests and more, that are specific to DT workflow └── tt-workflow/ // feature/functionality tests, recovery tests and more, that are specific to TT workflow └── tim-workflow/ // feature/functionality tests, recovery tests and more, that are specific to TIM workflow └── scale/ // tests that collect measurements on Voltha performance └── software-upgrades/ // tests for ONOS Applications, Voltha Components and ONU Software Upgrades └── dmi-interface/ // tests for device management interface └── openonu-go-adapter/ // tests state for ONU Go adapter with different operator workflows, different constellations and tech profiles └── bbsim/ // tests that are specific to BBSim functionality └── pm-data/ // tests for Open ONU Go adapter PM data └── memory-leak/ // tests to catch memory leak in Voltha components └── libraries // shared test keywords (functions) across various test suites └── variables // shared variables across various test suites └── scripts // shared python scripts across various test suites
An easy way to bring up VOLTHA + BBSim for testing is by using the helm-charts as described in voltha-helm-charts README. To set up a minimal environment, first install Docker.
NOTE: Please make sure you are able to run the docker command (your user is in the
dockergroup)
Then you can follow all the instructions in the voltha-helm-charts README.
Assuming that you have brought up VOLTHA as described above, you can run a simple E2E "sanity" test as follows:
git clone https://github.com/opencord/voltha-system-tests make -C voltha-system-tests sanity-single-kind
The tests generate three report files in voltha-system-tests/tests/functional (output.xml, report.html, log.html). View the report.html page in a browser to analyze the results. If you're running on a remote system, you can start a web server with python3 -m http.server.
To run the sanity tests for the DT Workflow, use sanity-kind-dt as the make target.
git clone https://github.com/opencord/voltha-system-tests make -C voltha-system-tests sanity-kind-dt
The tests generate three report files in voltha-system-tests/tests/dt-workflow/ (output.xml, report.html, log.html). View the report.html page in a browser to analyze the results. If you're running on a remote system, you can start a web server with python3 -m http.server.
To run the sanity tests for the TT Workflow, use sanity-kind-tt as the make target.
git clone https://github.com/opencord/voltha-system-tests make -C voltha-system-tests sanity-kind-tt
The tests generate three report files in voltha-system-tests/tests/tt-workflow/ (output.xml, report.html, log.html). View the report.html page in a browser to analyze the results. If you're running on a remote system, you can start a web server with python3 -m http.server.
To run the sanity tests for the TIM Workflow, use sanity-kind-tim as the make target.
git clone https://github.com/opencord/voltha-system-tests make -C voltha-system-tests sanity-kind-tim
The tests generate three report files in voltha-system-tests/tests/tim-workflow/ (output.xml, report.html, log.html). View the report.html page in a browser to analyze the results. If you're running on a remote system, you can start a web server with python3 -m http.server.
Consider an example, make sanity-single-kind target for ATT, which is a shortcut that specifies a number of variables used by the tests:
ROBOT_FILE: The test suite file in tests/functional that will be invoked by robot.
ROBOT_MISC_ARGS: Robot arguments passed directly to robot, for example to specify which test cases to run. For some environments you may need to override some default variable settings for your environment. See variables.robot for the list of defaults.
ROBOT_CONFIG_FILE: The YAML pod deployment file used to drive the test. Examples are in the tests/data directory.
Assuming that a POD is available with all the required hardware and connections, we can deploy the POD by following the procedure in this section below.
Deploying POD can be either manual or automated using Jenkins job. You can install it manually by following the steps in voltha-helm-charts README
Note: please add -f PATH_TO_YOUR_K8S_CONFIG_FILE to your helm commands with your Kubernetes configuration file. To create one please check this example.
The dataplane tests evaluate whether bandwidth and tech profiles are working as expected. These tests will only run on a physical pod. In order to run them it is required to manually install some additional software on the POD hosts that emulate the RG and BNG.
On the RG hosts:
iperf3 version 3.7, available here: https://software.es.net/iperf/jq: sudo apt install jqmausezahn: sudo apt install netsniff-ngsudo with no password: tcpdump, mausezahn, pkillOn the BNG host:
iperf3 version 3.7, available here: https://software.es.net/iperf/iperf3 in server mode in the background: iperf3 --server -Djq: sudo apt install jqmausezahn: sudo apt install netsniff-ngsudo with no password: tcpdump, mausezahn, pkillIn the POD's deployment config file, specify login information for the BNG host using the noroot_ip, noroot_user, and noroot_pass options. See the Tucson pod's config for an example.
All functional test cases for ATT are placed under functional folder. For DT, the folder is dt-workflow. Similarly for TT, it is tt-workflow.
Voltha_PODTests.robot and other similar robot files consist of functional testcases that can be run on a physical POD.
Each robot testcase has a description in the Documentation section.
The same suite of tests can be run on any POD because parameters needed for the test are written in .yaml file. Instead of hardcoding the POD specific variables in the test case, tests rely on a separate configuration file which describes the POD setup. This .yaml file contains details like the ONUs, OLT, nodes etc.
To create a configuration file for your POD, check this example
Input data are stored in the data folder. Few examples of input data could be, test specific SADIS configurations, tech profiles etc. Please give appropriate file names to the input files.
To trigger tests on the physical POD, say for ATT,
git clone https://github.com/opencord/voltha-system-tests cd voltha-system-tests make voltha-test ROBOT_FILE="Voltha_PODTests.robot" ROBOT_CONFIG_FILE="<PATH_TO_YOUR_POD_CONFIGURATION_FILE>"
Note: PATH_TO_YOUR_POD_CONFIGURATION_FILE should point to the YAML file that describes your POD setup.
Scenarios in each test suite can be associated with a Tag, using which a particular scenario can be invoked during test execution. As an example to execute only one testcase with tag test1 from the test suite you can run:
make voltha-test ROBOT_MISC_ARGS="-i test1" ROBOT_FILE="Voltha_PODTests.robot" ROBOT_CONFIG_FILE="<PATH_TO_YOUR_POD_CONFIGURATION_FILE>"
Tests should be written in RobotFramework as they need to be integrated with Jenkins test jobs. Libraries can be written in python or RobotFramework. Most additions should be done by adding keywords to the libraries, then calling these keywords from the tests. Consult a guide on how to write good Robot framework tests.
The cord-robot package provides a number of useful keywords for writing VOLTHA tests. See this link for information on how to import the library into a Robot test suite. The cord-robot package version is specified in the requirements.txt file. The package is automatically installed into the Python virtualenv set up by the Makefile.
Make sure that make lint check passes, which runs robotframework-lint on any new code that is created. The goal of the linter is to ensure that code is well formatted and structured, and that test suites are of a reasonable size. Lint can fail for a variety of reasons, usually related to formatting.
If you have trouble with the line length check, try the following:
If you get a Line Length related problem, you can continue lines between keywords with the ... operator - see the robot documentation for more information.
If it's an issue with a long shell invocation that uses a pipeline to filter output, try to see if you could use built-in Robot functionality for string or JSON manipulation, rather than using shell tools like sed, awk, or jq.
If you absolutely must use a long shell command, it can be stored in a string that is split over multiple lines with the Catenate Keyword before it's run.