[6.3] Reinstall capsule on the same node after remove (BZ: 1327442)

[ldjebran]

cover https://bugzilla.redhat.com/show_bug.cgi?id=1327442

(sat-6.3.0) dlezz@elysion:~/projects/robottelo-fork$ pytest -v tests/foreman/cli/test_capsule_installer.py::CapsuleInstallerTestCase::test_positive_reinstall_on_same_node_after_remove
================================================= test session starts ==================================================
platform linux2 -- Python 2.7.13, pytest-3.1.3, py-1.4.34, pluggy-0.4.0 -- /home/dlezz/.pyenv/versions/sat-6.3.0/bin/python2.7
cachedir: .cache
rootdir: /home/dlezz/projects/robottelo-fork, inifile:
plugins: xdist-1.20.0, services-1.2.1, mock-1.6.2, forked-0.2, cov-2.5.1
collected 1 item 
2017-10-19 14:34:08 - conftest - DEBUG - Deselect of WONTFIX BZs is disabled in settings


tests/foreman/cli/test_capsule_installer.py::CapsuleInstallerTestCase::test_positive_reinstall_on_same_node_after_remove <- robottelo/decorators/__init__.py PASSED

============================================= 1 passed in 5900.26 seconds ==============================================

[codecov]

Codecov Report

Merging #5388 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master    #5388   +/-   ##
=======================================
  Coverage   60.67%   60.67%           
=======================================
  Files          34       34           
  Lines        3697     3697           
=======================================
  Hits         2243     2243           
  Misses       1454     1454

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 9c67322...a9efad0. Read the comment docs.

[oshtaier]

ACK

[pondrejk]

looking at the bz, wouldn't it be less expensive to check if all appropriate files were deleted after remove, rather than installing the capsule again?

[pondrejk]

Or differently, we could assert that configs were removed after katello-remove finishes, this way if something slips, we wouldn't have to wait for failing capsule-installer to find out, does it make sense?

[abalakh]

ACK pending @pondrejk suggestion - assert config files are absent before reinstalling the capsule. And also your test takes around 99 minutes, pls move it to tier4 instead of tier3

[ldjebran]

The BZ say that the install failed because of existing configs and certs.

The main BZ complains is about install failure after remove. So install after remove is our goal to track

@abalakh @pondrejk this is not a test coverage of katello-remove
If it was the case we have to check all the packages, files and certs was removed we have also to check that all services are not working , but this is not the case

This is a test coverage of install after remove command

[ldjebran]

@abalakh @pondrejk even if we check that all the configs where removed, we have no guaranty that the capsule will reinstall successfully , so the only way to ensure that, is to reinstall it for real

[abalakh]

@ldjebran you probably didn't get me right. I'm not asking about removing reinstall part of the test (actually i'm against it, that's a nice coverage which we hadn't before), but proposing to add one extra assertion. If configs were not removed then reinstall won't succeed so there's no point in proceeding with scenario. Consider it smth like our self.assertEqual(result.return_code, 0) asserts - which aren't the things we need to verify in our tests but are saving us time since there's no point in proceeding when we know already that one of key commands has failed.

[abalakh]

Ok, nevermind, as the test is called test_positive_reinstall_on_same_node_after_remove maybe you have a point, it's not a bad idea to test reinstall no matter what's happening on lower level, were or were not some configs removed. Basically, even if they were not removed but capsule installation succeeded - we're good to mark the test with 'PASS'.

ACK