Poo#19230 Stabilize patch before migration tests on PPC by typing slowly #2976
Conversation
rwx788
changed the title
rwx788
force-pushed
the
poo#19230_patch_before_migration
branch
from
6da9974
to
637d081
Compare
lib/opensusebasetest.pm
Outdated
| @@ -332,7 +332,7 @@ sub wait_boot { | |||
| elsif (match_has_tag('inst-bootmenu')) { | |||
| # assuming the cursor is on 'installation' by default and 'boot from | |||
| # harddisk' is above | |||
| send_key_until_needlematch 'inst-bootmenu-boot-harddisk', 'up'; | |||
| # send_key_until_needlematch 'inst-bootmenu-boot-harddisk', 'up'; | |||
There was a problem hiding this comment.
that's probably only for testing, right?
There was a problem hiding this comment.
Yes, sorry, this was accidentally committed.
| my ($type_slow) = @_; | ||
| my $yast_scc_cmd = "yast2 scc; echo yast-scc-done-\$?- > /dev/$serialdev\n"; | ||
| #Due to ninja keys on PPC, may be required to type command slowlier, see poo#19230 | ||
| if ($type_slow) { |
There was a problem hiding this comment.
do you have an idea why we should type slower in this specific case and not in general? There is also the variable VNC_TYPING_LIMIT to limit overall typing speed. And we have set this for ppc64le already so maybe we really need to lower the limit. Unless it was about "QA-Power-…" behaving weird, which is another progress ticket
There was a problem hiding this comment.
Yes, tests on malbec are much more stable. Exactly this line failed couple of times due to redirection symbol, which requires shift key, other strings seem to work fine. This is same case when alt key is pressed and sometimes not released before. In general this change has to be reverted once it is fixed in framework, as I believe it should work even if worker is slow.
There was a problem hiding this comment.
we still have no idea if this is broken in qemu, power kernel or firmware - and that it's machine specific does not really give good an impression on the architecture :(
But just as okurz said, changing our tests because one worker is broken doesn't sound like a fair deal. Set the VNC_TYPING_LIMIT to really, really slow on that one worker.
There was a problem hiding this comment.
Agreed, changing VNC_TYPING_LIMIT is better idea as may help with other test failures like one here: https://progress.opensuse.org/issues/19266 https://progress.opensuse.org/issues/19414 which never happened on malbec.
| @@ -20,7 +20,7 @@ sub is_smt_or_module_tests { | |||
| } | |||
|
|
|||
| sub system_prepare() { | |||
| select_console 'root-console'; | |||
| select_console('root-console'); | |||
There was a problem hiding this comment.
why change? I think we should avoid reworking unrelated sections when it's not adding real value
There was a problem hiding this comment.
Thanks for the comment, missed this change also, will revert it.
rwx788
changed the title
rwx788
force-pushed
the
poo#19230_patch_before_migration
branch
from
637d081
to
1ce5510
Compare
rwx788
changed the title
rwx788
changed the title
In couple of runs redirection key was not properly typed, in other runs shift key was still pressed and wrong serial device was used. Issue occurs only on PowerPC and is not reproducible in every run. By typing unstable command slower we should gain more stability.
rwx788
force-pushed
the
poo#19230_patch_before_migration
branch
from
1ce5510
to
ffd108c
Compare
rwx788
changed the title
|
Trying with VNC_TYPING_LIMIT option and finding out real root cause instead of doing workaround for single worker. |
In couple of runs redirection key was not properly typed, in other runs shift key was still pressed and wrong serial device was used. Issue occurs only on PowerPC and is not reproducible in every run. By typing unstable command slower we should gain more stability.
Progress ticket: https://progress.opensuse.org/issues/19230
Verification run: http://10.160.66.147/tests/216#step/patch_before_migration/16