action #61019

new needle can not download during development mode

Added by coolgw almost 3 years ago. Updated almost 3 years ago.

Concrete Bugs
Target version:
Start date:
Due date:
% Done:


Estimated time:


1) open development mode
2) after assert specific screen timeout, then open needle edit, add new needle accept_license-license-agreement-s390-11sp4-20191213
3) back to develpment mode and resume the test, but error report show needle_dir not exist and download for new needle failed

Detail log you can check:

[2019-12-13T04:08:17.179 CET] [debug] isotovideo: pausing test execution on timeout as requested at installation-accept_license
[2019-12-13T04:11:51.348 CET] [debug] isotovideo: test execution will be resumed
Use of uninitialized value $needle_dir in concatenation (.) or string at /usr/lib/os-autoinst/OpenQA/Isotovideo/ line 76.
Use of uninitialized value $needle_dir in concatenation (.) or string at /usr/lib/os-autoinst/OpenQA/Isotovideo/ line 76.
[2019-12-13T04:11:51.353 CET] [debug] download new needle: => /accept_license-license-agreement-s390-11sp4-20191213.json
[2019-12-13T04:11:51.381 CET] [debug] unable to store download under /accept_license-license-agreement-s390-11sp4-20191213.json
[2019-12-13T04:11:51.381 CET] [debug] download new needle: => /accept_license-license-agreement-s390-11sp4-20191213.png
[2019-12-13T04:11:51.403 CET] [debug] unable to store download under /accept_license-license-agreement-s390-11sp4-20191213.png
[2019-12-13T04:11:51.498 CET] [debug] init needles from /var/lib/openqa/cache/
[2019-12-13T04:11:53.331 CET] [debug] loaded 10361 needles
[2019-12-13T04:11:53.816 CET] [debug] no match: 29.9s, best candidate: accept_license-license-agreement--20190510 (0.00)
[2019-12-13T04:11:54.797 CET] [debug] no change: 28.9s
[2019-12-13T04:11:55.794 CET] [debug] no change: 27.9s
[2019-12-13T04:11:56.798 CET] [debug] no change: 26.9s


#1 Updated by okurz almost 3 years ago

  • Category set to Concrete Bugs
  • Priority changed from Normal to High

Thanks for the report. Sounds like a regression.

From recent osd deployment the changes are:


$ git log1 --no-merges 5daf6345..163a29fa
c6972328 Remove "use Devel::Cover" from 07-commands.t
c3f4d0ab t/22-svirt.t: add open_serial_console_via_ssh test
fcc97ef1 Makefile: Check for spellcheck requirements
45148918 sshXtermVt: check for SSH port available instead of a simple ping to check host availability
3b138d1e svirt: open_serial_console_via_ssh - doc and cleanup
f2165ae1 Add new test/cover targets and run tests with prove
023870a8 Fix warning on failed load_snapshot command
d9d6eb27 backend_vars: Add infos for generalhw ssh/noVNC and flash cmd
6a13fc98 generalhw: allow to use real serial with ssh/noVNC mode
5d660025 sshXtermVt: wait that SUT is alive on network before starting ssh in xterm Use SSH_XTERM_WAIT_SUT_ALIVE_TIMEOUT to adjust wait time, default 120 s
766aeda5 generalhw: Add flash command support and handle optionnal args for cmd
4f542c86 generalhw: Enable SSH access when VNC is not selected
af27dcf0 generalhw: truncate serial file on start

and for openQA:

$ git log1 --no-merges 22c773f5c..76dde5b6f
9c061cf0e Clarify overlap between states CANCELLED and DONE
4d890418f Improve consistency when dealing with job states/results
e96a23d1f Dependency cron 191208000552
54d9ea600 openqa-clone-custom-git-refspec: Add support for comma-separated job list
167e80b6e openqa-clone-custom-git-refspec: Fix github token authentication by removing quotes
0a7b61932 Update Mojolicious to 8.27 (#2576)
3ccc1c06b Fix height of embedded logfile
df3630548 Use the correct new port in cache service monitoring tests
3ecf80b86 Allow monitoring of Minion jobs on every worker
097c8f50a travis_test: Add missing dictionary for os-autoinst

no obvious hits. Maybe dependency updates?

Sounds more likely that eccf76b3 Allow relative "productdir" and "needledir" relative to "casedir" could have caused this.

#2 Updated by Xiaojing_liu almost 3 years ago

I discussed this problem with coolgw, the use a variable $needle::needles_dir to get the needle directory, but we did not find the needles_dir, there is another variable in calls $needledir. So should change the needles_dir to needledir?

#3 Updated by okurz almost 3 years ago

First I would like to understand what has happened. 8e359ba9 has introduced the use of the variable but this was already in 2019-10. If it means the developer mode is broken since then and we did not realize I think it's also very important that we improve our tests. Regarding variable names I would prefer "$needle_dir" over "$needledir".

#4 Updated by okurz almost 3 years ago

  • Status changed from New to In Progress
  • Assignee set to okurz
  • Target version set to Current Sprint

still looking for a way to have a unit test that also shows the problem.

#5 Updated by okurz almost 3 years ago

  • Status changed from In Progress to Feedback

merged. A followup is as an improvement to prevent the problem about referencing an in existant needle dir variable in the future with getter/setter methods.

#6 Updated by okurz almost 3 years ago

  • Status changed from Feedback to Resolved
  • Target version changed from Current Sprint to Done

both last mentioned PRs are merged and deployed to both o3 and osd meanwhile. I did not observe further problems regarding needle download when last trying on o3 so I guess it's ok now.

Also available in: Atom PDF