I agree more information is better, however it is not obvious where that information will come from.
- Test case run result history Do we keep records of old runs in Colorado for functest & yardsticks? If we do, let’s link them up. If not, we can always re-run these tests on the frozen Colorado release and produce these results. Are we regularly running them now? Also note, the Colorado release is frozen, test cases are frozen, so this spot info may not be as relevant as it appears. However I agree it’ll become more informational and valuable with longer history. - Source code of test cases Do we have a link to source code repos in openstack, ODL/ONOS, etc upstreams? Can someone involved in CI/CD pitch in? Wenjing From: SULLIVAN, BRYAN L [mailto:bs3...@att.com] Sent: Friday, January 27, 2017 6:20 AM To: Wenjing Chu <wenjing....@huawei.com>; Pierre Lynch <ply...@ixiacom.com>; Jose Lausuch <jose.laus...@ericsson.com> Cc: TECH-DISCUSS OPNFV <opnfv-tech-discuss@lists.opnfv.org> Subject: RE: [opnfv-tech-discuss] [dovetail]L3VPN for dovetail area More inline. Thanks, Bryan Sullivan | AT&T From: Wenjing Chu [mailto:wenjing....@huawei.com] Sent: Thursday, January 26, 2017 12:30 PM To: SULLIVAN, BRYAN L <bs3...@att.com<mailto:bs3...@att.com>>; Pierre Lynch <ply...@ixiacom.com<mailto:ply...@ixiacom.com>>; Jose Lausuch <jose.laus...@ericsson.com<mailto:jose.laus...@ericsson.com>> Cc: TECH-DISCUSS OPNFV <opnfv-tech-discuss@lists.opnfv.org<mailto:opnfv-tech-discuss@lists.opnfv.org>> Subject: RE: [opnfv-tech-discuss] [dovetail]L3VPN for dovetail area Hi Bryan Hope my inline responses are still readable … Thanks. Wenjing From: SULLIVAN, BRYAN L [mailto:bs3...@att.com] Sent: Thursday, January 26, 2017 8:29 AM To: Wenjing Chu <wenjing....@huawei.com<mailto:wenjing....@huawei.com>>; Pierre Lynch <ply...@ixiacom.com<mailto:ply...@ixiacom.com>>; Jose Lausuch <jose.laus...@ericsson.com<mailto:jose.laus...@ericsson.com>> Cc: TECH-DISCUSS OPNFV <opnfv-tech-discuss@lists.opnfv.org<mailto:opnfv-tech-discuss@lists.opnfv.org>> Subject: RE: [opnfv-tech-discuss] [dovetail]L3VPN for dovetail area More comments inline. Thanks, Bryan Sullivan | AT&T From: Wenjing Chu [mailto:wenjing....@huawei.com] Sent: Wednesday, January 25, 2017 3:00 PM To: SULLIVAN, BRYAN L <bs3...@att.com<mailto:bs3...@att.com>>; Pierre Lynch <ply...@ixiacom.com<mailto:ply...@ixiacom.com>>; Jose Lausuch <jose.laus...@ericsson.com<mailto:jose.laus...@ericsson.com>> Cc: TECH-DISCUSS OPNFV <opnfv-tech-discuss@lists.opnfv.org<mailto:opnfv-tech-discuss@lists.opnfv.org>> Subject: RE: [opnfv-tech-discuss] [dovetail]L3VPN for dovetail area Thanks Bryan. See my response inline below. Wenjing From: SULLIVAN, BRYAN L [mailto:bs3...@att.com] Sent: Wednesday, January 25, 2017 11:32 AM To: Wenjing Chu <wenjing....@huawei.com<mailto:wenjing....@huawei.com>>; Pierre Lynch <ply...@ixiacom.com<mailto:ply...@ixiacom.com>>; Jose Lausuch <jose.laus...@ericsson.com<mailto:jose.laus...@ericsson.com>> Cc: TECH-DISCUSS OPNFV <opnfv-tech-discuss@lists.opnfv.org<mailto:opnfv-tech-discuss@lists.opnfv.org>> Subject: RE: [opnfv-tech-discuss] [dovetail]L3VPN for dovetail area I posted some comments in gerrit. Here are the main points I think we need alignment on: 1) All proposed dovetail included tests will be added one-by-one, in a separate commit. Please follow the gerrit tickets below and see if you can follow through. Test cases are organized into two levels for convenience: test areas and test cases. There will be one commit for each test case, and one commit for each test area (which includes a lot of test cases that are related to a function area). The test case commit says we are good on how that test case is implemented. The test area commit says we agree the test case ought to be included. Clear enough? [bryan] A test reference should not be added to a test area until the test has been approved (e.g. verified by more than one committer/reviewer). I see one commit with about 40 tests referenced. Have all these been verified? [wenjing] the act of “adding a test case to a test area” is reflected as adding a single line in the compliance_set.yml file. Using your example, I can add 40 tests in a patch, meaning, I “propose” (based on the wiki discussion in the past) that these 40 cases be included. The submitter is sending this proposal out for review. If you have opinion about any of these, or simply want more explanation/clarification, please comment on the gerrit patch. Based on those comments, the submitter can modify the patch to reflect updated view and resubmit, until we are good and approve with +1 (or disapprove with -1). Hope that is clear. These are patch reviews, to be approved. Not yet. I’m not sure what you mean by “verified”. If you mean if the software is tested, I think the answer is yes. If you mean if the test case has been approved, no, a patch email is precisely asking you to review. [bryan] It will be much more complicated to approve a block of tests for inclusion, rather than test-by-test, unless it is clarified that as a group they have been passing (e.g by reference to Jenkins jobs where the tests have been successfully running as a group). A single large commit with a bunch of text strings with no explanation as to what they are, where they have been tested, etc, is not the way we should manage the dovetail testcase list. 2) The commit will include a link to the details of the test case (script or otherwise what I would use to run the test for myself) You can trace down to the source step by step, e.g. from test area to test case, then to functest or yardstick, and/or to openstack or other upstream eventually the source code in that upstream project. To test run it, you would need a test environment/pod. I would think that running dovetail tool, specifying the individual test cases you’d want to run, and examining the results probably a good way to go. Maybe good to write down a “how-to” cheat-sheet for this? [bryan] Not sure how that answered the comment. Rather than having to search for something that relates to the test case reference, it would be good for the commit message (for test cases) to contain a URL reference to the test source. That’s what I was referring to. We need to simplify the effort of reviewers, to encourage more active reviews and second-opinions/testing on the tests. Re the test environment, that’s no problem the test should clarify what is needed and how to run it. Having an environment to do so is clear, and should be clarified by the test anyway. [wenjing] I’ll be happy to try a simpler way for us to review upstream code, so if anyone has better idea, please suggest here. But let me first make sure I understand what you are asking. If a piece of code is written in Dovetail project, yes, your gerrit patch will show everything. If we are only referencing a test case in Functest, you will only see the reference line of source in Dovetail. I believe we’ve provided in that line a handle to locate the source in Functest (which can be a lot more line or files). If then Functest references another test in Openstack or ODL, then the source info is buried in CI/CD process and ultimately out of OPNFV. We may not even have a ready copy to reference other than may be github clones. Without full CI/CD integration between projects, it’s not obvious how one can precisely locate them. Let me ask around how this can be done. But before I embark on this, is that what you are looking for? Anything else I missed? I will look into the how-to doc on running individual test cases. [bryan] I believe the request is clear; we need direct evidence that each test is passing in CI/CD. Having to go hunt for (1) information about what the test is about; (2) evidence that it is a regular successful test in CI/CD (meaning that it is executed successfully on a regular basis), are both ineffective ways to manage the approved tests. If we are going to use Gerrit for the review process (which I think is a bit too low-level of a tool – I prefer JIRA), then the reviews need to be as self-sufficient as possible, with at most direct links to the test cases as proposed for inclusion, and the evidence of regular successful runs in CI/CD. 3) All tests need to be working under at least one scenario, and the more scenarios that have been validated (either explicitly or implicitly), the higher priority the test should get. “Implicit” means that a test validated on a basic scenario (e.g. nosdn) is implicitly validated on other scenarios for that installer. But explicit validation is of course best. Thanks for highlighting the implicit cases: more “implicit” is “better”, because it means something works more “universally” rather than relying on special cases. I would caution on the “more scenario” metrics again because it does not necessarily mean “larger applicability”. Sometimes it does, sometimes it doesn’t. Also note the fact that we ought not be counting non-generic scenarios as the same as generic ones. So let’s not be too numerical about it, the criteria should be about the larger applicability scope. I made this point in one my earlier emails as well going through the scenarios in Colorado. [bryan] Sure, more scenarios tested does not necessarily mean more broadly validated. But at least for the basic installer scenarios, validation across more installers should bump up the priority of the test. [wenjing] agreed. 4) The reviewers may require that they be able to duplicate the test validation before commit merge. Please refer to 2) and see if you need anything else. /// Thanks, Bryan Sullivan | AT&T From: opnfv-tech-discuss-boun...@lists.opnfv.org<mailto:opnfv-tech-discuss-boun...@lists.opnfv.org> [mailto:opnfv-tech-discuss-boun...@lists.opnfv.org] On Behalf Of Wenjing Chu Sent: Wednesday, January 25, 2017 11:26 AM To: Pierre Lynch <ply...@ixiacom.com<mailto:ply...@ixiacom.com>>; Jose Lausuch <jose.laus...@ericsson.com<mailto:jose.laus...@ericsson.com>> Cc: TECH-DISCUSS OPNFV <opnfv-tech-discuss@lists.opnfv.org<mailto:opnfv-tech-discuss@lists.opnfv.org>> Subject: Re: [opnfv-tech-discuss] [dovetail]L3VPN for dovetail area The process that we may have already being informally following is as follows. We work towards having consensus on majority of areas that arise within the dovetail project. If there are open questions that we can’t resolve, we could gather the relevant info and bring that to TSC for decision. In the TSC review, dovetail will present the proposed plan out of dovetail, plus potentially open issues, and ask for (a) approval of the proposal (b) determination of open questions, if any. Does this sound like a good process to follow? On the topic of scenario cleanup, the Dovetail team has been voicing that opinion for a long time, and so I applaud and strongly support the effort to separate general vs specific scenarios, and it will help Dovetail tremendously going forward. However, also keep in mind that that work is slated for D and E releases. It unfortunately can’t help in the immediate task for us for C release target. To join in the detailed review effort, please note that review of test areas and test cases are based on Jira and Gerrit. For example, These are for test areas: (the file is compliance_set.yml) https://gerrit.opnfv.org/gerrit/27493 https://gerrit.opnfv.org/gerrit/27219 And here is an example of a test case within a test area: https://gerrit.opnfv.org/gerrit/27221 These Gerrit links are also posted on wiki for convenience: https://wiki.opnfv.org/display/dovetail/Dovetail+Test+Areas+and+Test+Cases However it’s a bit slow to refresh there since it is a manual process. I would recommend you get on gerrit still. We are at the beginning of the review process so it’s not late. General level questions or specific topics can of course still be done in mailing list or on meetings, but try to stay on gerrit as much as you can. Let us know if you have any feedback. Thanks. Regards Wenjing From: opnfv-tech-discuss-boun...@lists.opnfv.org<mailto:opnfv-tech-discuss-boun...@lists.opnfv.org> [mailto:opnfv-tech-discuss-boun...@lists.opnfv.org] On Behalf Of Pierre Lynch Sent: Wednesday, January 25, 2017 8:44 AM To: Jose Lausuch <jose.laus...@ericsson.com<mailto:jose.laus...@ericsson.com>> Cc: TECH-DISCUSS OPNFV <opnfv-tech-discuss@lists.opnfv.org<mailto:opnfv-tech-discuss@lists.opnfv.org>> Subject: Re: [opnfv-tech-discuss] [dovetail]L3VPN for dovetail area IMHO, getting agreement on what the scope of testing will be (features, etc) should be pretty urgent. How should we go about it? Agree within the Dovetail team, then run it to the TSC to get their blessing? Should we consolidate this process with the current ongoing discussion on scenario consolidation, which lead to the idea of generic versus specific scenarios? Dovetail would include generic scenarios, while specific scenarios would be excluded from Dovetail? It would provide uniformity…. I would expect that determining what’s in and what’s out could be a delicate process. Thanks, Pierre On Jan 25, 2017, at 7:18 AM, Jose Lausuch <jose.laus...@ericsson.com<mailto:jose.laus...@ericsson.com>> wrote: Thanks Chris, that makes things clearer. But still, it is a broad statement and difficult to measure. I guess and as you say, the TSC has the final word when approving features to be verified/certified in Dovetail with existing tests. From functional prospective, I can just provide and overview about how the tests were behaving when releasing Colorado. Regards, Jose From: Christopher Price Sent: Wednesday, January 25, 2017 14:17 PM To: Jose Lausuch; Tianhongbo; Tim Irnich Cc: 'TECH-DISCUSS OPNFV' Subject: Re: [opnfv-tech-discuss] [dovetail]L3VPN for dovetail area Hi Jose, The intent of this statement is that we should not attempt to establish compliance tests on features or capabilities that are unique to a very specific configuration or composition of components. The statement is intended to mean that we should focus our efforts on compliance on “generally available” or “community relevant” use cases and features. Again, I am not able to accurately articulate what that means or how to measure it, as such we have a somewhat obtuse statement in the documentation. This should be seen as a guideline to be applied by the development, testing and dovetail teams around expectations for compliance testing. It would be eventually ratified or judged by the TSC as they have the final say on the tests that are approved for compliance validation for a given dovetail release. Does that help? I do believe we should formalize and commit our governance into a repo and have the TSC cast an approving eye over it as well for good form. Then, if nothing else, we would have a more consistent view of our intention and needed approach. / Chris From: <opnfv-tech-discuss-boun...@lists.opnfv.org<mailto:opnfv-tech-discuss-boun...@lists.opnfv.org>> on behalf of Jose Angel Lausuch Sales <jose.laus...@ericsson.com<mailto:jose.laus...@ericsson.com>> Date: Wednesday, 25 January 2017 at 11:51 To: Tianhongbo <hongbo.tianhon...@huawei.com<mailto:hongbo.tianhon...@huawei.com>>, Tim Irnich <tim.irn...@ericsson.com<mailto:tim.irn...@ericsson.com>> Cc: TECH-DISCUSS OPNFV <opnfv-tech-discuss@lists.opnfv.org<mailto:opnfv-tech-discuss@lists.opnfv.org>> Subject: Re: [opnfv-tech-discuss] [dovetail]L3VPN for dovetail area Hi Hongbo, Test cases must pass on OPNFV reference deployments • Tests must not require a specific NFVi platform composition or installation tool Can you please explain what this statement exactly means? By “installation tool” are we talking about the installers we have or a specific and different tool to install a certain feature? Adding Tim, who is the SDNVPN PTL. Thanks, Jose From: Tianhongbo [mailto:hongbo.tianhon...@huawei.com] Sent: Wednesday, January 25, 2017 01:46 AM To: Jose Lausuch Cc: 'TECH-DISCUSS OPNFV' Subject: [dovetail]L3VPN for dovetail area Hi Jose: As you mentioned, there will be discussion about the more detail of the L3VPN with L3VPN team to check if the L3VPN can be included in the dovetail area now. There are some requirements from the dovetail wiki page: https://wiki.opnfv.org/display/dovetail/Dovetail+Test+Case+Requirements<https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwiki.opnfv.org%2Fdisplay%2Fdovetail%2FDovetail%2BTest%2BCase%2BRequirements&data=01%7C01%7Cplynch%40ixiacom.com%7Ca899fc5c2cac41cc1b5e08d445357c28%7C069fd614e3f843728e18cd06724a9b23%7C0&sdata=%2FCI%2BUD%2F5T2kB12q%2Bv1Be2QLxpb3uvv34yrJ%2B2uUyyrA%3D&reserved=0> Look forward to your reply. Best regards hongbo _______________________________________________ opnfv-tech-discuss mailing list opnfv-tech-discuss@lists.opnfv.org<mailto:opnfv-tech-discuss@lists.opnfv.org> https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.opnfv.org%2Fmailman%2Flistinfo%2Fopnfv-tech-discuss&data=01%7C01%7Cplynch%40ixiacom.com%7Ca899fc5c2cac41cc1b5e08d445357c28%7C069fd614e3f843728e18cd06724a9b23%7C0&sdata=tEw44iTeTSA%2FTtVV877qqLFhR%2FR8mwD2MbB1OBUZCP0%3D&reserved=0
_______________________________________________ opnfv-tech-discuss mailing list opnfv-tech-discuss@lists.opnfv.org https://lists.opnfv.org/mailman/listinfo/opnfv-tech-discuss
