our biggest "customer" uses nis+autofs, so we should have a test for that :)
- AC1: We have a NIS server+client test in openSUSE TW
- AC2: Same as AC1 but for openSUSE Leap 15.2
- AC3: Test suites are enabled in development job group first and then moved to the main job group
- AC4: Same as AC1 but for SLE 15 DONE: https://openqa.suse.de/tests/1404255
- AC5: Same as AC1 but for SLE 12 SP4 DONE: https://openqa.suse.de/tests/1408217#step/nis_client/43
- Crosscheck existing NIS tests against manual expectations
- If necessary add new tests / extend existing tests
- Ensure the supported testsuites are called in all products+versions mentioned above
SLE 15 tests can be used as a base:
#5 Updated by RBrownSUSE over 5 years ago
- % Done changed from 100 to 80
While I really like the test and it mostly works (75% of the problems we've seen are due to changes in SLE12 SP2 that are NOTHING to do with the test) we've still had to remove the yast2_nis test from main.pm for the following reasons
- it only works on the suse.de network - no one can test anything in Prague or Bejing with it present
- it's reliant on nis.suse.de - this prevents the test being used by openSUSE, which is always a goal if it can be done
- while it is important, it gets in the way of staging. It probably warrants either it's own scenario, or we're going to have to start implementing something like the extra_tests_ scenarios which openSUSE uses for stuff like this
Zaoliang, please make the following changes
- Please make it so this test does not rely on nis.suse.de.
- This might mean making two tests, one to setup a NIS/NFS server, and then connecting the NIS client to it using an improved version of yast2_nis.pm
- I'd accept localhost testing, but this might be an excuse to experiment with openQA's multi-machine testing - Denis is our expect.
This will make the test more flexible in where we put it and how we put it..once we get there I'll hopefully have made a decision about how we put it back in the sle main.pm :)
#9 Updated by dzedro almost 5 years ago
- Status changed from New to In Progress
YaST nis code is pretty old, bugs below are blocking progress in openQA test
Existing openQA test https://openqa.suse.de/tests/627400
#11 Updated by okurz almost 5 years ago
- Status changed from In Progress to Feedback
I just checked all bugs you mentioned. There are all in state NEW and I don't see that any immediate action is necessary by you so I guess we are blocked here by others. Setting to feedback to resemble the blocked state, I hope you don't mind
#17 Updated by okurz over 4 years ago
I also checked the bugs
Not any useful progress at all.
But: dzedro: I could not find a recent job example. Possibly because we are restructuring the "extra tests" to be more aligned with the structure on o3. When you say the test is "blocked": Is NIS unusable? If this is the case I think we would need to escalate. If NIS works but the test has different assumptions then it would need to be adjusted to reflect reality. How feasible would this be?
#29 Updated by okurz almost 4 years ago
- Subject changed from [sle][functional][opensuse]nis+autofs to [sle][functional][hard][opensuse]nis+autofs
- Description updated (diff)
- Assignee changed from sebchlad to okurz
https://openqa.suse.de/tests/1408217#step/nis_client/43 shows the client for SLE 12 SP4 and it actually checks nicely the expectations, e.g. exported home directory over NFS from test user on NIS.
I will try to harmonize the test plan regarding the test modules "nis_server" and "nis_client". "yast2_nis" relies on nis.suse.de, not much we can do here. The challenge though is that currently there are no multi-machine capable workers in the o3 infrastructure. I see the following options:
- openvswitch+multi-machine o3 workers
- rewrite tests be one-host only
I asked in #opensuse-factory what we can do about opt. 1.
#37 Updated by okurz over 2 years ago
- Assignee changed from okurz to riafarov
Move to new QSF-y PO after I moved to the "tools"-team. I mainly checked the subject line so in individual instances you might not agree to take it over completely into QSF-y. Feel free to reassign to me or someone else in this case. Thanks.
#53 Updated by JRivrain over 1 year ago
I haven't really been working on this, as I worked on some bug reports. What I see so far is I have to rewrite parts of the test to work in opensuse, as opensuse desktop is using NetworkManager. So there are 2 options:
1 - rewrite the network part to work also with NM
2 - use a textmode image as it uses wicked also on opensuse, when not desktop. but then, it also needs a bit of rewrite, as the test is made for desktop
#55 Updated by JRivrain over 1 year ago
- Status changed from In Progress to Blocked
https://github.com/os-autoinst/os-autoinst-distri-opensuse/pull/9646. This works, but, ticket is blocked because of https://progress.opensuse.org/issues/64195
#56 Updated by JRivrain over 1 year ago
- Status changed from Blocked to In Progress
The other issue is hopefully fixed for good.
Now, I found this problem, posted on yast's IRC:
Hello, I am testing yast nis_server module on TW, and I see that there are different files mapped in TW and in sle, see TW: http://waaa-amazing.suse.cz/tests/11777#step/nis_server/48 vs SLE http://waaa-amazing.suse.cz/tests/11773#step/nis_server/51. Does somebody here know why is that so ?
And another: we have a problem in our test, if we compare the screenshot vs the needle here we see that the needle matches, but not all mapped files are selected. So that's a false positive that stays un-detected. Maybe we can address this in a different ticket.
EDIT: for the latter, actually if we check the video, it's working as expected. So that's just that the needle matches too early at 98%.
#57 Updated by JRivrain over 1 year ago
Apparently, the difference between files is due to https://github.com/yast/yast-nis-server/blob/ed65ef561f14a16705e2ce2e5fd4cee5c8c9dc37/src/include/nis_server/maps.rb#L88, so it's normal. But then how to expect how many of them we should select ?
#59 Updated by JRivrain over 1 year ago
I reported that bug: https://bugzilla.suse.com/show_bug.cgi?id=1167589. basically the test does not work in TW. We could work around this, but I also found another problem: the already existing workarounds are outdated, we have one soft-failure about firewalld that refers a closed bug, see closure comment here: https://bugzilla.suse.com/show_bug.cgi?id=1083487#c36.
#60 Updated by JRivrain over 1 year ago
Added a workaround for the reported bug, removed outdated workarounds that were for closed bugs, or slee<15-sp2 ( as it seems like we (qa-sle-y) are the only ones now to use this test suite)
#61 Updated by JRivrain over 1 year ago
- Checklist changed from [x] SLE, [ ] Leap, [ ] TW to [x] SLE, [x] TW
- Status changed from In Progress to Feedback
PR merged, test enabled in TW development group. Discussed with Leap PO, we agreed that in today's context it did not make much sense to test it on Leap. Putting it on feedback until we see if it all works fine.