1 .. This work is licensed under a Creative Commons Attribution 4.0 International License.
2 .. SPDX-License-Identifier: CC-BY-4.0
4 ***********************************************************
5 NOTE - This file will be updated during the Lakelse Release
6 ***********************************************************
8 ***********************
9 Testing developer guide
10 ***********************
21 The OPNFV testing ecosystem is wide.
23 The goal of this guide consists in providing some guidelines for new developers
24 involved in test areas.
26 For the description of the ecosystem, see `[DEV1]`_.
32 There are several ways to join test projects as a developer. In fact you may:
34 * Develop new test cases
36 * Develop tooling (reporting, dashboards, graphs, middleware,...)
37 * Troubleshoot results
38 * Post-process results
40 These different tasks may be done within a specific project or as a shared
41 resource accross the different projects.
43 If you develop new test cases, the best practice is to contribute upstream as
44 much as possible. You may contact the testing group to know which project - in
45 OPNFV or upstream - would be the best place to host the test cases. Such
46 contributions are usually directly connected to a specific project, more details
47 can be found in the user guides of the testing projects.
49 Each OPNFV testing project provides test cases and the framework to manage them.
50 As a developer, you can obviously contribute to them. The developer guide of
51 the testing projects shall indicate the procedure to follow.
53 Tooling may be specific to a project or generic to all the projects. For
54 specific tooling, please report to the test project user guide. The tooling used
55 by several test projects will be detailed in this document.
57 The best event to meet the testing community is probably the plugfest. Such an
58 event is organized after each release. Most of the test projects are present.
60 The summit is also a good opportunity to meet most of the actors `[DEV4]`_.
62 Be involved in the testing group
63 ================================
65 The testing group is a self organized working group. The OPNFV projects dealing
66 with testing are invited to participate in order to elaborate and consolidate a
67 consistant test strategy (test case definition, scope of projects, resources for
68 long duration, documentation, ...) and align tooling or best practices.
70 A weekly meeting is organized, the agenda may be amended by any participant.
71 2 slots have been defined (US/Europe and APAC). Agendas and minutes are public.
72 See `[DEV3]`_ for details.
73 The testing group IRC channel is #opnfv-testperf
78 All the test projects do not have the same maturity and/or number of
79 contributors. The nature of the test projects may be also different. The
80 following best practices may not be acurate for all the projects and are only
81 indicative. Contact the testing group for further details.
87 Most of the projects have a similar structure, which can be defined as follows::
101 | `-- Dockerfile.aarch64.patch
116 Test projects are installing tools and triggering tests. When it is possible it
117 is recommended to implement an API in order to perform the different actions.
119 Each test project should be able to expose and consume APIs from other test
120 projects. This pseudo micro service approach should allow a flexible use of
121 the different projects and reduce the risk of overlapping. In fact if project A
122 provides an API to deploy a traffic generator, it is better to reuse it rather
123 than implementing a new way to deploy it. This approach has not been implemented
124 yet but the prerequisites consiting in exposing and API has already been done by
125 several test projects.
130 Most of the test projects provide a docker as deliverable. Once connected, it is
131 possible to prepare the environement and run tests through a CLI.
136 Dockerization has been introduced in Brahmaputra and adopted by most of the test
137 projects. Docker containers are pulled on the jumphost of OPNFV POD.
138 <TODO Jose/Mark/Alec>
143 It is recommended to control the quality of the code of the testing projects,
144 and more precisely to implement some verifications before any merge:
148 * unit tests (python 2.7)
149 * unit tests (python 3.5)
151 The code of the test project must be covered by unit tests. The coverage
152 shall be reasonable and not decrease when adding new features to the framework.
153 The use of tox is recommended.
154 It is possible to implement strict rules (no decrease of pylint score, unit
155 test coverages) on critical python classes.
161 Several test projects integrate third party tooling for code quality check
162 and/or traffic generation. Some of the tools can be listed as follows:
164 +---------------+----------------------+------------------------------------+
165 | Project | Tool | Comments |
166 +===============+======================+====================================+
167 | Bottlenecks | TODO | |
168 +---------------+----------------------+------------------------------------+
169 | Functest | Tempest | OpenStack test tooling |
170 | | Rally | OpenStack test tooling |
171 | | Refstack | OpenStack test tooling |
172 | | RobotFramework | Used for ODL tests |
173 +---------------+----------------------+------------------------------------+
174 | QTIP | Unixbench | |
179 +---------------+----------------------+------------------------------------+
180 | Storperf | TODO | |
181 +---------------+----------------------+------------------------------------+
183 +---------------+----------------------+------------------------------------+
184 | Yardstick | Moongen | Traffic generator |
185 | | Trex | Traffic generator |
186 | | Pktgen | Traffic generator |
187 | | IxLoad, IxNet | Traffic generator |
189 | | Unixbench | Compute |
190 | | RAMSpeed | Compute |
191 | | LMBench | Compute |
192 | | Iperf3 | Network |
193 | | Netperf | Network |
194 | | Pktgen-DPDK | Network |
195 | | Testpmd | Network |
196 | | L2fwd | Network |
198 | | Bonnie++ | Storage |
199 +---------------+----------------------+------------------------------------+
202 ======================================
203 Testing group configuration parameters
204 ======================================
209 The testing group defined several categories also known as tiers. These
210 categories can be used to group test suites.
212 +----------------+-------------------------------------------------------------+
213 | Category | Description |
214 +================+=============================================================+
215 | Healthcheck | Simple and quick healthcheck tests case |
216 +----------------+-------------------------------------------------------------+
217 | Smoke | Set of smoke test cases/suites to validate the release |
218 +----------------+-------------------------------------------------------------+
219 | Features | Test cases that validate a specific feature on top of OPNFV.|
220 | | Those come from Feature projects and need a bit of support |
221 | | for integration |
222 +----------------+-------------------------------------------------------------+
223 | Components | Tests on a specific component (e.g. OpenStack, OVS, DPDK,..)|
224 | | It may extend smoke tests |
225 +----------------+-------------------------------------------------------------+
226 | Performance | Performance qualification |
227 +----------------+-------------------------------------------------------------+
228 | VNF | Test cases related to deploy an open source VNF including |
229 | | an orchestrator |
230 +----------------+-------------------------------------------------------------+
231 | Stress | Stress and robustness tests |
232 +----------------+-------------------------------------------------------------+
233 | In Service | In service testing |
234 +----------------+-------------------------------------------------------------+
239 The domains deal with the technical scope of the tests. It shall correspond to
240 domains defined for the certification program:
254 One of the goals of the testing working group is to identify the poorly covered
255 areas and avoid testing overlap.
256 Ideally based on the declaration of the test cases, through the tags, domains
257 and tier fields, it shall be possible to create heuristic maps.
259 =============================================
260 Reliability, Stress and Long Duration Testing
261 =============================================
263 Resiliency of NFV refers to the ability of the NFV framework to limit disruption
264 and return to normal or at a minimum acceptable service delivery level in the
265 face of a fault, failure, or an event that disrupts the normal operation
268 **Reliability** testing evaluates the ability of SUT to recover in face of fault,
269 failure or disrupts in normal operation or simply the ability of SUT absorbing
272 Reliability tests use different forms of faults as stimulus, and the test must
273 measure the reaction in terms of the outage time or impairments to transmission.
275 **Stress testing** involves producing excess load as stimulus, and the test
276 must measure the reaction in terms of unexpected outages or (more likely)
277 impairments to transmission.
279 These kinds of "load" will cause "disruption" which could be easily found in
280 system logs. It is the purpose to raise such "load" to evaluate the SUT if it
281 could provide an acceptable level of service or level of confidence during such
282 circumstances. In Danube and Euphrates, we only considered the stress test with
283 excess load over OPNFV Platform.
285 In Danube, Bottlenecks and Yardstick project jointly implemented 2 stress tests
286 (concurrently create/destroy VM pairs and do ping, system throughput limit)
287 while Bottlenecks acts as the load manager calling yardstick to execute each
288 test iteration. These tests are designed to test for breaking points and provide
289 level of confidence of the system to users. Summary of the test cases are listed
290 in the following addresses:
292 * https://wiki.opnfv.org/display/bottlenecks/Stress+Testing+over+OPNFV+Platform
293 * https://wiki.opnfv.org/download/attachments/2926539/Testing%20over%20Long%20Duration%20POD.pptx?version=2&modificationDate=1502943821000&api=v2
295 **Stress test cases** for OPNFV Euphrates (OS Ocata) release can be seen as
296 extension/enhancement of those in D release. These tests are located in
297 Bottlenecks/Yardstick repo (Bottlenecks as load manager while Yardstick execute
298 each test iteration):
300 * VNF scale out/up tests (also plan to measure storage usage simultaneously): https://wiki.opnfv.org/pages/viewpage.action?pageId=12390101
301 * Life-cycle event with throughputs (measure NFVI to support concurrent
302 network usage from different VM pairs):
303 https://wiki.opnfv.org/display/DEV/Intern+Project%3A+Baseline+Stress+Test+Case+for+Bottlenecks+E+Release
305 In OPNFV E release, we also plan to do **long duration testing** over OS Ocata.
306 A separate CI pipe testing OPNFV XCI (OSA) is proposed to accomplish the job.
307 We have applied specific pod for the testing.
308 Proposals and details are listed below:
310 * https://wiki.opnfv.org/display/testing/Euphrates+Testing+needs
311 * https://wiki.opnfv.org/download/attachments/2926539/testing%20evolution%20v1_4.pptx?version=1&modificationDate=1503937629000&api=v2
312 * https://wiki.opnfv.org/download/attachments/2926539/Testing%20over%20Long%20Duration%20POD.pptx?version=2&modificationDate=1502943821000&api=v2
314 The long duration testing is supposed to be started when OPNFV E release is
316 A simple monitoring module for these tests is also planned to be added:
317 https://wiki.opnfv.org/display/DEV/Intern+Project%3A+Monitoring+Stress+Testing+for+Bottlenecks+E+Release
324 Where can I find information on the different test projects?
325 ============================================================
326 On http://docs.opnfv.org! A section is dedicated to the testing projects. You
327 will find the overview of the ecosystem and the links to the project documents.
329 Another source is the testing wiki on https://wiki.opnfv.org/display/testing
331 You may also contact the testing group on the IRC channel #opnfv-testperf or by
332 mail at test-wg AT lists.opnfv.org (testing group) or opnfv-tech-discuss AT
333 lists.opnfv.org (generic technical discussions).
336 How can I contribute to a test project?
337 =======================================
338 As any project, the best solution is to contact the project. The project
339 members with their email address can be found under
340 https://git.opnfv.org/<project>/tree/INFO
342 You may also send a mail to the testing mailing list or use the IRC channel
346 Where can I find hardware resources?
347 ====================================
348 You should discuss this topic with the project you are working with. If you need
349 access to an OPNFV community POD, it is possible to contact the infrastructure
350 group. Depending on your needs (scenario/installer/tooling), it should be
351 possible to find free time slots on one OPNFV community POD from the Pharos
352 federation. Create a JIRA ticket to describe your needs on
353 https://jira.opnfv.org/projects/INFRA.
354 You must already be an OPNFV contributor. See
355 https://wiki.opnfv.org/display/DEV/Developer+Getting+Started.
357 Please note that lots of projects have their own "how to contribute" or
358 "get started" page on the OPNFV wiki.
361 How do I integrate my tests in CI?
362 ==================================
363 It shall be discussed directly with the project you are working with. It is
364 done through jenkins jobs calling testing project files but the way to onboard
365 cases differ from one project to another.
367 How to declare my tests in the test Database?
368 =============================================
369 If you have access to the test API swagger (access granted to contributors), you
370 may use the swagger interface of the test API to declare your project.
371 The URL is http://testresults.opnfv.org/test/swagger/spec.html.
373 .. figure:: ../../../images/swaggerUI.png
375 :alt: Testing Group Test API swagger
377 Click on *Spec*, the list of available methods must be displayed.
379 .. figure:: ../../../images/API-operations.png
381 :alt: Testing Group Test API swagger
383 For the declaration of a new project use the POST /api/v1/projects method.
385 For the declaration of new test cases in an existing project, use the POST
386 /api/v1/projects/{project_name}/cases method
388 .. figure:: ../../../images/CreateCase.png
390 :alt: Testing group declare new test case
392 How to push your results into the Test Database?
393 ================================================
395 The test database is used to collect test results. By default it is
396 enabled only for CI tests from Production CI pods.
398 Please note that it is possible to create your own local database.
400 A dedicated database is for instance created for each plugfest.
402 The architecture and associated API is described in previous chapter.
403 If you want to push your results from CI, you just have to call the API
404 at the end of your script.
406 You can also reuse a python function defined in functest_utils.py `[DEV2]`_
409 Where can I find the documentation on the test API?
410 ===================================================
412 The Test API is now documented in this document (see sections above).
413 You may also find autogenerated documentation in
414 http://artifacts.opnfv.org/releng/docs/testapi.html
415 A web protal is also under construction for certification at
416 http://testresults.opnfv.org/test/#/
418 I have tests, to which category should I declare them?
419 ======================================================
422 The main ambiguity could be between features and VNF.
423 In fact sometimes you have to spawn VMs to demonstrate the capabilities of the
424 feature you introduced.
425 We recommend to declare your test in the feature category.
427 VNF category is really dedicated to test including:
429 * creation of resources
430 * deployement of an orchestrator/VNFM
431 * deployment of the VNF
435 The goal is not to study a particular feature on the infrastructure but to have
436 a whole end to end test of a VNF automatically deployed in CI.
437 Moreover VNF are run in weekly jobs (one a week), feature tests are in daily
438 jobs and use to get a scenario score.
440 Where are the logs of CI runs?
441 ==============================
443 Logs and configuration files can be pushed to artifact server from the CI under
444 http://artifacts.opnfv.org/<project name>
451 `[DEV1]`_: OPNFV Testing Ecosystem
453 `[DEV2]`_: Python code sample to push results into the Database
455 `[DEV3]`_: Testing group wiki page
457 `[DEV4]`_: Conversation with the testing community, OPNFV Beijing Summit
459 `[DEV5]`_: GS NFV 003
461 .. _`[DEV1]`: http://docs.opnfv.org/en/latest/testing/ecosystem/index.html
462 .. _`[DEV2]`: https://git.opnfv.org/functest/tree/functest/utils/functest_utils.py#176
463 .. _`[DEV3]`: https://wiki.opnfv.org/display/meetings/Test+Working+Group+Weekly+Meeting
464 .. _`[DEV4]`: https://www.youtube.com/watch?v=f9VAUdEqHoA
465 .. _`[DEV5]`: http://www.etsi.org/deliver/etsi_gs/NFV/001_099/003/01.01.01_60/gs_NFV003v010101p.pdf
467 IRC support chan: #opnfv-testperf