Hi

Good try.

Could this be run by command line without CI?

summary report are good dimensions, on the other hand, it hides some info.

in tempest, minimum/api/scenario/service tests are different category. We 
usually to use percentage for the same  test case pass rate by running it with 
number of times.

/MatthewLi

发件人: [email protected] 
[mailto:[email protected]] 代表 Jose Lausuch
发送时间: 2016年7月29日 19:09
收件人: [email protected]
主题: [opnfv-tech-discuss] Functest feedback in Jenkins

Hi,

In order to provide a better user experience and useful feedback from the daily 
runs in Jenkins, these features have been implemented in Functest recently:

“Blocking flag” for each test case:
For each test case executed in Functest, there is a flag called “blocking” that 
can be set to True or False. If the test fails and the flag is set ‘true’, 
Functest will stop the execution of the rest of the tests in the daily loop 
exiting with error. If the test fails but the flag is set to false, it will 
mark the Job as failed, but it will allow the execution of the remaining test 
cases.
https://git.opnfv.org/cgit/functest/tree/ci/testcases.yaml
The purpose of this is to allow the critical test cases like the 
healthcheck/vping be a “gate” or a filter i.e. if you can’t create VMs, why 
should you bother in testing sfc or any other feature? Some other tests like 
Tempest, will allow the execution of the preceding tests even though if there 
is a failure.


Summary report:
Some of you gave us some feedback about the confusing output given by Functest 
in Jekins if a job has failed. This has been improved a lot, providing clear 
output messages to understand what is happening. But now you don’t even need to 
go through all that output, you just need to scroll  down and at the bottom of 
the page you will see the “Functest report” to check what has worked/failed 
along with some other useful information.
Example here: 
https://build.opnfv.org/ci/job/functest-fuel-baremetal-daily-master/261/console


Black listing tests in Tempest:
If you know that in a certain scenario/installer, a tempest test case fails, 
you can now blacklist the test case in this yaml:
https://git.opnfv.org/cgit/functest/tree/testcases/OpenStack/tempest/custom_tests/blacklist.txt
For example, for bgpvpn we have deactivated the tests that deal with floating 
ips due to lacking support for floating IPs in ODL VPNService (which will be 
fixed in the future).
So, if you know that a known bug will make a test case fail, you can add the 
test in this list and it will be skipped.
NOTE: this is not meant to avoid running tests that fail without a reason. It 
should be justified (known bug, feature lack of support, …).


You’ll see a lot of red in Jenkins, and that’s because Functest has become a 
bit more strict when giving the final status. It will only return blue if all 
the tests succeeded according to the defined success criteria, not only if the 
execution of the test script succeeded (even though the test failed).
For example:
https://build.opnfv.org/ci/job/functest-fuel-baremetal-daily-master/261/console
https://build.opnfv.org/ci/job/functest-fuel-baremetal-daily-master/260/console
The exit status is -1 because there was >0 failures in Tempest-smoke, and the 
success criteria for Tempest-smoke is 100%.

This strictness has been changed like this after some feedback collected from 
the community, in particular from the installer teams.



Thanks,

JOSE LAUSUCH
Senior Systems Designer
Ericsson

_______________________________________________
opnfv-tech-discuss mailing list
[email protected]
https://lists.opnfv.org/mailman/listinfo/opnfv-tech-discuss

Reply via email to