gNMI is becoming the network management protocol of choice for Model-Driven Telemetry. Regardless how well the gNMI Specifications may have been defined, there will still be requirements in it that are open for interpretation. To ensure consistency in gNMI telemetry server implementations between vendors and products, interoperability testing will help to identify areas of the implementations that are not compliant with the gNMI Specifications.
This project is intended as a tool to help with automated testing of the telemetry capabilities of target devices. It uses a Robot Framework (a generic open source automation framework) to implement a range of test suites and test cases. These try to identify capabilities of a device to service typical telemetry related gNMI requests.
We require Python3 for running the tests. All the python commands in following text assume that this binary resolves to a (fairly recent) python3 interpreter. Following is an example of supported (not necessarily lowest required) version:
$ python --version
Python 3.8.10See Robot's installation instructions for details on how to install the framework. To verify you have a robot installed, you can run robot --version shell command. As of writing this document, current version is:
$ robot --version
Robot Framework 6.0.2 (Python 3.8.10 on linux)Actual test-runs require that supplementary gNMI client code (located in sibling repository gnmi-tools, primarily confd_gnmi_client.py module) can be run successfully. See the gnmi-tools' README.md for requirements if you run into some issues.
Please note that the referred README file contains a lot of extra information/details not needed for the test-runs, and we don't imply going through it whole if not needed/interested.
You can generate the "specification" of all the test suites/test cases from the existing robot files without need of a live target device to test. You can do this using Robot's supporting tool called testdoc.
Assuming you have all the requirements installed, just execute following shell command from within the current working directory (the one where this README.md file is located):
python -m robot.testdoc ./ TestDocumentation.htmlThis creates a single HTML file with specified name - that includes descriptions and details of all the test suites and test-cases from the project's .robot files.
It has a structure of a collapsed tree nodes that you can open to see specific test-suites and/or test-cases hierarchy, and read their names, descriptions, etc.
If needed, see "python -m robot.testdoc --help" for possible output configuration.
To run any tests, you need a target device configuration file first.
When a test run is executed and finished, it creates few files by default (unless modified by robot command parameters).
report.html - summary of all the tests suites/cases and their results;log.html file for details...log.html - detailed log for all the steps executed in test run⚠ - Please note that log files will include your username and plaintext password configured for connecting to the target device (as described in further section on
.yamlfiles).
Whether you run tests against "real" device, or the "ConfD adapter" (see further), you need properly set Python environment for tests to execute correctly.
TODO - think of some executor script to save user from having to declare PYTHONPATH?
For our case, this implies setting the PYTHONPATH variable for underlying python code to find all the required items spread across various directories of the project. Assuming you run the tests from current directory, you need to have the following set for tests to run:
PYTHONPATH=../gnmi-tools/src:./:./General_gNMIeither exported in current environment, or passed as env. variable when invoking robot commands mentioned further...
To run the test cases against live target device, you need to write up your configuration file (e.g. "MY.yaml"). See adapter.yaml for an example of variables that need to be included in the file and their description. Create your own .yaml file that you can use in next step for executing the tests.
You may want to run only subset of tests. See robot's --help output for details on customization the test-run.
We recommend primarily the --include and/or --exclude options to allow selecting specific test-cases depending on existing test "tags". See the report.html/log.html or actual .robot files for list of existing tags.
A good starting is to run the tests with "sanity" tag to verify basic operation/connectivity. You can pass the YAML file as a parameter in command below to run the test suites:
robot --variablefile MY.yaml --include sanity ./"Successfully" executed (irrespective of actual test results) test run results in output into console like:
==============================================================================
Testtool :: Telemetry testing suite for a brief verification of the device'...
==============================================================================
Testtool.gNMI Interface :: Test suite dedicated to verification of generic ...
==============================================================================
Testtool.gNMI Interface.Capabilities :: Generic device agnostic test suite ...
==============================================================================
Device sends CapabilityResponse (any) :: Device can respond with `... | PASS |
------------------------------------------------------------------------------
Mandatory JSON is advertised as supported encoding :: Get the list... | FAIL |
[ JSON_IETF | ASCII | PROTO ] does not contain value 'JSON'.
------------------------------------------------------------------------------
[ WARN ] Mandatory JSON (as per gNMI specification) not declared as supported
Supported encodings include JSON or JSON_IETF :: Test if the devic... | PASS |
------------------------------------------------------------------------------
Device should support some models :: Check that the ``supported_mo... | PASS |
------------------------------------------------------------------------------
...output continued...
Output: .../output.xml
Log: .../log.html
Report: .../report.htmlYour specific output may vary, depending on how the set of existing test suites/test cases changes in time.
This results in creation of previously mentioned log files of interest - report.html and log.html.
You can run tests without having any target device, e.g. to see an example test outputs/logs. You can do this by using the so-called "ConfD gNMI adapter" as a target device.
ConfD gNMI adapter serves as a proof-of-concept implementation of gNMI enabled device. It utilizes the ConfD to run and service the model-specific requests for some of OpenConfig standard models. It includes a simple gNMI server written in Python to "proxy" the incoming gNMI requests into the ConfD database. It can also in so-called "demo" mode without actual ConfD demon required/running.
For details on the tools supporting these ROBOT test suites, you can see the gnmi-tools's README file.
Previously mentioned adapter.yaml file already contains all the variables settings needed for this use-case.
To run the gNMI Adapter demo as a "backend" for these test runs, please see the gnmi-tool's README file's section.
Quick command list to copy & test before digging to referred document:
start gNMI server proxy:
with ConfD available (adapter API mode):
make -C ../gnmi-tools stop clean all start
../gnmi-tools/src/confd_gnmi_server.py -t confd --insecureor without ConfD (adapter DEMO mode):
make -C ../gnmi-tools clean gnmi_proto
../gnmi-tools/src/confd_gnmi_server.py -t demo --insecurerun tests in other console:
PYTHONPATH=../gnmi-tools/src:./:./General_gNMI robot --variablefile adapter.yaml --variablefile interfaces.yaml --variablefile defaults.yaml --include sanity ./