Functest Release Notes

1. License

OPNFV Colorado release note for Functest Docs (c) by Morgan Richomme (Orange)

OPNFV Colorado release note for Functest Docs are licensed under a Creative Commons Attribution 4.0 International License. You should have received a copy of the license along with this. If not, see <>.

2. OPNFV Colorado1.0 release note for Functest

2.1. Abstract

This document describes the release note of Functest project.

2.2. Version history

Date Ver. Author Comment
2016-08-17 1.0.0 Morgan Richomme (Orange) Functest for Colorado release

2.3. OPNFV Colorado Release

Functest deals with functional testing of the OPNFV solution. It includes test cases developed within the project and test cases developed in other OPNFV projects and other upstream communities.

The internal test cases are:

  • healthcheck
  • vPing ssh
  • vPing userdata
  • Tempest Smoke Serial
  • Rally Sanity
  • ODL
  • Tempest full parallel
  • Rally full
  • vIMS

The OPNFV projects integrated into Functest framework for automation are:

  • bgpvpn
  • Copper
  • Doctor
  • Domino
  • Moon
  • Multisite
  • ONOS-sfc
  • ODL-sfc
  • Parser
  • Promise
  • Security scan

The validation of a scenario requires a subset of these tests depending on the installer and the scenario.

The 3 last internal test cases (tempest full parallel, Rally full and vIMS) are not considered for scenario validation.

2.4. Release Data

Project functest
Repo/tag colorado.1.0
Release designation Colorado base release
Release date September 22 2016
Purpose of the delivery Colorado base release

2.6. Version change

2.6.1. Feature evolution

  • refactoring of ODL functional tests (with upstream modifications)
  • refactoring of testapi (update, swagger documentation, dockerization)
  • jenkins logs improvement
  • update integration of Doctor, Promise and SDNVPN projects
  • split Tempest and rally into 2 different tests: smoke and full
  • vIMS test suite integration
  • adoption of Kibana for dashboarding

2.6.2. New features

  • Functest CLI to prepare and run the tests
  • creation of the healthcheck test case
  • support new scenarios (ocl, odl_l2-sfc, onos-sfc, lxd, moon, fdio, multisite)
  • integration of new OPNFV feature projects (copper, domino, multisite,

moon, parser, onos-sfc, odl-sfc, security scan)

  • introduction of test tiers in functest framework
  • automatic reporting
  • introduction of a jenkins summary table
  • support of ARM architecture

2.7. Scenario Matrix

For Colorado 1.0, Functest was tested on the following scenarios (if not precised, the scenario is a HA scenario):

Scenario Apex Compass Fuel Joid
nosdn X X X X
nosdn-noha     X X
kvm     X  
kvm-noha     X  
lxd       X
lxd-noha       X
ovs-noha (dpdk)     X  
odl_l2 X X X X
odl_l2-noha     X  
odl_l2-bgpvpn X   X  
odl_l2-bgpvpn-noha     X  
odl_l2-fdio-noha X      
odl_l2-moon   X    
odl_l2-sfc-noha X   X  
odl_l3 X X X  
odl_l3-noha     X  
onos   X X X
onos-noha     X  
onos-sfc   X X X
onos-sfc-noha     X  
multisite     X  

In Colorado, the functional tests have been sliced in 6 different categories:

Category Description
healthcheck Basic OpenStack commands
smoke vPings, Tempest and rally smoke tests
sdn_suites Specific SDN feature tests
features OPNFV feature project functional test suites
openstack Advanced, long duration OpenStack tests (Tempest and Rally full suite). Each test may last several hours
vnf Complex scenarios dealing with orchestration, VNF deployment and tests (e.g. vIMS)

For the scenario validation, we consider only the healthcheck, smoke, sdn_suites and features categories. These tests are run systematically in the CI daily loops.

Success criteria have been defined for these test cases, they can be PASS/FAIL or a success rate may be declared (100%, > 90%) All the criteria, as well as the test dependencies are declared in the ci/testcases.yaml file.

The scoring for the Colorado release per installer can be described as follows.

The scoring is an indicator showing how many feature project test suites have been integrated on the scenario.

The scoring equals the number of tests * succesful iteration of each test [0-3]. The scoring method is described in

e.g. apex/odl_l2-nofeature-ha tests = vping_ssh+vping_userdata+tempest+rally+odl+doctor+copper Scoring = 21/21 = 7 * 3

By default, if not specified, the scenarios are HA. HA means OpenStack High Availability (main services). Note that not all VIM (e.g. OpenStack) services are deployed in HA mode, as that depends upon support of the specific service for HA deployment. For example, in the Colorado release, the Congress service is deployed in non-HA mode even for HA OPNFV scenarios, as explicit support for HA operation has not yet been verified. See the release notes of the installers for more details.

2.7.1. Apex

Scenario Scoring Success rate Results
nosdn 17/18 95% apex-res-174
odl_l2 21/21 100% apex-res-175
odl_l3 15/18 83% apex-res-176
odl_l2-bgpvpn 14/18 78% apex-res-235
odl_l2-fdio-noha 12/15 80% apex-res-6
odl_l2-sfc-noha 18/24 75% apex-res-217

2.7.2. Compass

Scenario Scoring Success rate Results
nosdn 12/12 100% compass-res-55
odl_l2 15/15 100% compass-res-59
odl_l3 9/12 75% compass-res-73
odl_l2-moon 15/18 83% compass-res-567
onos-ha 15/15 100% compass-res-77
onos-sfc-ha 17/18 95% compass-res-76

Note: all the Compass tests for Colorado have been executed on virtual environment. Bare metal resources were used for Master branch.

2.7.3. Fuel

Scenario Scoring Success rate Results
nosdn 18/18 100% fuel-res-129
nosdn-noha 15/15 100% fuel-res-154
nosdn-kvm 18/18 100% fuel-res-128
nosdn-kvm-noha 15/15 100% fuel-res-161
nosdn-ovs 12/18 67%* fuel-res-213
nosdn-ovs-noha 15/15 100% fuel-res-162
odl_l2 21/21 100% fuel-res-123
odl_l2-noha 17/18 94% fuel-res-155
odl_l2-bgpvpn 14/18 78% fuel-res-119
odl_l2-bgpvpn-noha 14/15 93% fuel-res-160
odl_l2-sfc-noha 6/21 29% fuel-res-219
odl_l3 15/18 83% fuel-res-115
odl_l3-noha 12/15 80% fuel-res-164
onos 20/21 95% fuel-res-117
onos-noha 18/18 100% fuel-res-166
onos-sfc 24/24 100% fuel-res-124
onos-sfc-noha 21/21 100% fuel-res-129
multisite N.R 100% fuel-res-8

*: all results passed, lacking iterations to reach the full score

Results of Functest on AArch64 Colorado 1.0

Scenario Scoring Success rate Results
nosdn 18/18 100% fuel-arm-res-29
odl_l2 21/21 100% fuel-arm-res-24
odl_l2-noha 17/18 94% fuel-arm-res-30
odl_l3 14/18 78% fuel-arm-res-28

2.7.4. Joid

Scenario Scoring Success rate Results
nosdn 18/18 100% joid-res-102
nosdn-noha 17/18 95% joid-res-93
nosdn-lxd 12/12 100% joid-res-104
nosdn-lxd-noha 12/12 100% joid-res-91
odl_l2 21/21 100% joid-res-103
onos 21/21 100% joid-res-99
onos-sfc 24/24 100% joid-res-97

It is highly recommended to install a json viewer in your browser (e.g.

You can get additional details through test logs on As no search engine is available on the OPNFV artifact web site you must retrieve the pod identifier on which the tests have been executed (see field pod in any of the results) then click on the selected POD and look for the date of the test you are interested in.

The reporting pages can be found at:

2.8. Colorado known restrictions/issues

Installer Scenario Issue
any odl_l3-* Tempest cases related to using floating IP addresses fail because of a known ODL bug. vPing_ssh test case is excluded for the same reason.
apex/fuel *-bgpvpn vPing_ssh (floating ips not supported) and vIMS excluded. Some Tempest cases related to floating ips also excluded. Some performance issues have been detected in this scenario (i.e. BGPVPN extension enabled) when running commands against the OpenStack APIs, thus Rally sanity test case has been disabled. Performance issues seem to be connected to the ODL version. It is planned to reintroduce Rally sanity in Colorado 2.0 with the adoption of ODL Boron release.
apex *-fdio Due to late integration, fdio decided to focus on mandatory tests and exclude feature tests (copper, doctor, security_scan) from its scenarios
compass moon First ODL test FAILS because ODL/Openstack federation done in moon is partial. Only MD-SAL is federated (not AD-SAL)
fuel any TestServerBasicOps test case skipped
fuel kvm vPing_ssh and vIMS excluded (metadata related scenarios)
fuel multisite random errors running multisite. A known bug in keystone mitaka, due to which memcache raises exception and keystone becomes unresponsive workaround consists in restarting memcache on server
joid any Tempest cases related to object storage excluded
joid domino Domino tests are skipped in CI. However the test case can be run by manually setting IS_IPandKEY_CONFIGURED=true after manually setting properly the IP addresses of the 3 Controller nodes in the configuration file /repos/domino/tests/

2.9. Test and installer/scenario dependencies

It is not always possible to run all the test cases on all the scenarios. The following table details the dependencies of the test cases per scenario. The scenario dependencies (installer or scenario) are detailed in

2.10. Test results

Test results are available in:

2.11. Open JIRA tickets

JIRA Description
FUNCTEST-446 Cleanup ODL-SFC output in Functest execution Impact on odl_l2-sfc scenarios
FUNCTEST-454 Cleanup failures when using HA networks in Neutron


Author:Morgan Richomme (

Build date: September 22, 2016