That other one didn't help, so let's try this - try and spot if
the spoke is in the unexpected state (the needle should only
match if the spoke is done processing and still in warning
state, it shouldn't match while the needle is still thinking)
and click through it again if so.
Well, that OCR needle isn't working out so great, as it seems
to match when it shouldn't:
https://openqa.fedoraproject.org/tests/119217#step/_graphical_wait_login/5
So let's try another approach. Ditch the OCR needle and have a
function for checking we're at a clean desktop. It does the
normal needle match, but if we're on GNOME, it also tries
hitting alt+f1 and seeing if we're at the overview; if so, it
hits alt+f1 again (to go back to the desktop) and returns.
IIRC disk_guided_empty is the only storage test that clicks
through INSTALLATION DESTINATION *really* fast, so let's try
adding a 2-second sleep to it to see if it works around the
'sometimes spoke shows as incomplete' bug that cropped up in
F26 and hasn't been fixed yet and tends to cause failures.
https://openqa.fedoraproject.org/tests/117131 shows a spurious
match of the anaconda warning_bar needle - it's matching in an
ad for LibreOffice. Add a bit of the grey above the bar, so
hopefully this won't happen any more.
RHBZ #1222413 was fixed long ago. This workaround is, I think,
the cause of openQA failures to run commands properly with an
extraneous '2' at the start of the command (e.g. 116864).
Summary:
As we're getting the Workstation dvd-ostree (OStree installer
image) built for Rawhide now, let's try testing it.
Test Plan:
Run the tests on a Rawhide compose that works and
has the image (e.g. 20170615.n.0). Check that new tests work
as expected and old tests are not adversely affected. A
corresponding diff for fedora_openqa will be coming to take
care of scheduling. Note that the tests will often soft fail
for now; this is intentional due to RHBZ#1193590, the bash
prompt for root is incorrect on ostree installs, so I have
added a needle that matches the incorrect prompt but which is
flagged as a workaround needle (so causing the test result to
be a soft fail).
Reviewers: jsedlak, jskladan
Reviewed By: jsedlak
Subscribers: tflink
Differential Revision: https://phab.qa.fedoraproject.org/D1211
The F26 background does not change with the time of day, so that
needle should serve as-is, it's just no use for Rawhide. For
Rawhide, let's try a needle with an OCR match on 'Activities',
plus a tiny match area for a pure white block (expected to be
matched in the user menu area), since os-autoinst requires all
needles to include a match area for some reason.
These are no use :/ The Rawhide default background is a 'changes
with the time of day' one, and with the transparent top bar,
that makes a proper needle match impossible to reliably achieve,
as there is not a single pixel of the 'clean desktop' which
will always be the same any more. We'll have to come up with
something else. I'm trying out the use of an OCR match now.
The GNOME top bar is now transparent in Rawhide, which broke the
'clean desktop' needle. It also means it'll break every time
the background changes, from now on :( Update the needle.
Summary:
This just adds the FreeIPA web UI and password change
test modules to the FreeIPA upgrade test (client end). It's
useful to check out these features too. We don't need to
separate these into separate jobs, as we're not trying to
fill out different matrix checkboxes here, we just want to
know whether everything works.
Test Plan:
Run the test, see that the modules work properly.
I was actually expecting this to fail given the issues with
the upgrade on the server end, but it seems to pass.
Reviewers: jsedlak, jskladan
Reviewed By: jsedlak
Subscribers: tflink
Differential Revision: https://phab.qa.fedoraproject.org/D1207
Summary:
This adds an upgrade variant of the FreeIPA tests, with only
the simplest client enrolment (sssd) for now. The server test
starts from the N-1 release and deploys the domain controller
role. The client test similarly starts from the N-1 release
and, when the server is deployed, enrols as a domain client.
Then the server upgrades itself, while the client waits (as the
server is its name server). Then the client upgrades itself,
while the server does some self-checks. The server then waits
for the client to do its checks before decommissioning itself,
as usual. So, summary: *deployment* of both server and client
occurs on N-1, then both are upgraded, then the actual *checks*
occur on N.
In my testing, this all more or less works, except the role
decommission step fails. This failure seems to be a genuine one
so far as I can tell; I intend to file a bug for it soon.
Test Plan:
Run the new tests, check they work. Run the existing
FreeIPA tests (both the compose and the update variants), check
they both behave the same.
Reviewers: jsedlak, jskladan
Reviewed By: jsedlak
Subscribers: tflink
Differential Revision: https://phab.qa.fedoraproject.org/D1204
Seems like the top '[patch]' link on kernel.org may now be a
raw patch file which the browser just displays, not a compressed
patch it offers to download. So tweak the needle so we should
click on the *bottom* link instead. openQA looks for matches
as close as possible to the location in the needle.
When running universal tests with a non-Server image, which is
not common but *does* happen (e.g. manually built test image),
the pre-selected filesystem in blivet-gui will be ext4 not xfs
(just as with custom partitioning). So we need xfs and ext4
variants of this needle. Renamed the XFS one for consistency.
Previously this module changed test1's password, so that it
would still be able to work even if the webui test module
failed (so test3/test4 didn't get created). But this means
that, for about 30 seconds, test1's password is 'loremipsum'
not 'batterystaple', and if one of the *other* client test
jobs happens to hit a point where it has to auth as test1
during the 30 seconds test1's password is different, it will
fail. This looks to be what happened to the join_kickstart
test the last few days - it failed because it tried to login
as test1 during the password change window.
By using the test3 user instead (which is only used by the
join_cockpit test, currently) we avoid this problem, at the
cost that the password_change module will always fail if the
webui module fails.
With the latest F26 base images, it seems like g-i-s fails to
run at first login. This is clearly some kind of bug somewhere
and I'll investigate it, but it shouldn't be causing the update
tests to fail - we can still validly run the tests with g-i-s
not running. So for now, adjust the _graphical_wait_login test
to tolerate this behaviour when running update tests.
Summary:
Loading the same module more than once *kinda* works, but it
shows up all kinds of funky in the openQA web interface. There's
a POO for this:
https://progress.opensuse.org/issues/10514
But it doesn't seem like it's going to be resolved immediately,
so in the mean time maybe we should avoid doing it so we don't
have to deal with the weirdness it produces in the web UI. So
here's a kinda icky hack that uses symlinks and stuff to load
multiple instances of 'the same' test module.
Test Plan:
Run an update test, look at how it looks in the web
UI and confirm it's a lot clearer and less buggy. Check there
aren't any bugs in the loading approach. This is deployed on stg
so you can look at it there.
Reviewers: jsedlak, jskladan
Reviewed By: jsedlak
Subscribers: tflink
Differential Revision: https://phab.qa.fedoraproject.org/D1186
It's not really a good idea to have the comments that explain
the test_flags in *every* test, because they can go stale and
then we either have to live with them being old or update them
all. Like, now. So let's just take 'em all out. There's always
a reference in the openQA and os-autoinst docs, and those get
updated faster.
More importantly, add the new `ignore_failure` flag to relevant
tests - all the tests that don't have the 'important' or
'fatal' flag at present. Upstream killed the 'important' flag
(making all tests 'important' by default), I got it replaced
with the 'ignore_failure' flag, we now need to explicitly mark
all modules we want the 'ignore_failure' behaviour for.
The way this was set up before, if `anaconda_main_hub` matched
immediately but some spoke was still in a 'processing' state,
it only had 30 seconds (default `assert_and_click` timeout) to
complete and allow the 'Begin Installation' button to appear.
It seems unnecessary to match on *both* needles, really, so
let's just give 300 seconds for the `begin_installation` needle
to appear. It's not going to appear on any other screen.
This problem caused a couple of spurious failures today -
https://openqa.fedoraproject.org/tests/77839 and
https://openqa.fedoraproject.org/tests/77858 - because they
took a bit too long for the INSTALLATION DESTINATION spoke to
clear.
