From Fedora Project Wiki

< FWN‎ | Beats

Revision as of 20:55, 2 November 2009 by Adamwill (talk | contribs) (create 200 qa beat (sorry it's late!))

QualityAssurance

In this section, we cover the activities of the QA team[1].

Contributing Writer: Adam Williamson

  1. ↑ http://fedoraproject.org/wiki/QA

Test Days

Last week's Test Day[1] was on internationalization (also known as i18n)[2]. We had a good turnout of testers who covered a wide variety of languages and input methods. In general many appear to be in good shape, but the testing turned up several issues in Bengali, Malayalam and a few other languages. This testing will help us to improve the implementation of these languages in future. Rui He provided a summary[3] of the event, including a list of all bugs filed.

No Test Day is planned for next week. If you would like to propose a main track Test Day for the Fedora 13 cycle, please contact the QA team via email or IRC, or file a ticket in QA Trac[4].

  1. ↑ http://fedoraproject.org/wiki/Test_Day:2009-10-29
  2. ↑ http://fedoraproject.org/wiki/I18N
  3. ↑ https://www.redhat.com/archives/fedora-test-list/2009-November/msg00038.html
  4. ↑ http://fedorahosted.org/fedora-qa/

Weekly meetings

The QA group weekly meeting[1] was held on 2009-10-26. The full log is available[2]. James Laska reported that he had renamed most of the Debugging pages[3] to follow the previously agreed-upon naming scheme. The only remaining page was KernelBugTriage, and he would check with kernel maintainers before renaming this one.

James Laska noted that Marcela Maslanova had written automated testing scripts for the previous week's Test Day[4], and this had produced a very positive experience. He asked the group to think about what future Test Days could potentially benefit from testing automation in this way.

Adam Williamson passed along a proposal from Milos Jakubicek that the QA and BugZappers group help with filing bugs on the remaining Fedora 12 packages with FTBFS (fails to build from scratch) issues. Jesse Keating pointed out that Matt Domsch has a script which tries to rebuild all of Rawhide and automatically files bugs on packages which fail, which he typically runs once per cycle. Jesse believed the fact that Milos is aware of several packages which fail to build but for which no bug report currently exists is a result of the fact that the list Milos is working from was generated a month after Matt's latest test run. The group agreed that Adam would ask Milos to clarify his proposal and see if it was still necessary in light of the existence of Matt's script.

Jóhann Guðmundsson presented his proposal for an automated test of non-U.S. locale installation, prompted by the significant bugs[5] [6] in the Beta with installations with different locale settings which were not caught by pre-release testing. He pointed out that implementing such a test would be relatively simple and involve only defining a non-U.S. locale in a kickstart file for an installation test run. The group agreed that this would be valuable testing and asked Jóhann to write it up into a test case that could be added to the installation test matrix and also potentially automated as part of future AutoQA development.

Will Woods and Kamil Paral reported on the progress of the AutoQA project. Kamil had made a blog post announcing rpmguard to the world[7]. He had received feedback from several people, including suggestions from Seth Vidal and Alexey Torkhov (whose feedback had prompted a ticket[8]). Kamil is now planning to work on integrating rpmguard into AutoQA with the help of the newly-implemented Koji watcher, which allows AutoQA to pick up - and potentially trigger tests upon - every new build which goes through Koji. Will briefly touched upon the future organization plan for all the AutoQA code, based around a library for the server-side parts such as watchers and another library for actual tests, along with separate configuration files for things like the relationships between Koji tags, so these configuration details can be separated from the main functional code. Will also noted that he had created a Python script for generating the current set of critical path packages[9]: simply running it generates the list as critpath.txt. He plans to have this integrated into the Rawhide compose process so that a daily updated critical path package list is always available at a static URL. Finally, Will noted that a public mailing list has been created for the AutoQA project, autoqa-devel[10]. James Laska noted in passing that the hardware for the production AutoQA instance was currently likely to be delivered on 2009-11-20.

James Laska reviewed upcoming events. He noted that preparation for the then-upcoming i18n Test Day[11] was well advanced, and asked for group members to help out with testing if they could. He trailed the then-upcoming second Fedora 12 blocker bug review day, which would take place on 2009-10-30, and Adam Williamson asked people to help by re-testing blocker bugs prior to the event and coming to the event to help walk the list.

The Bugzappers group weekly meeting[12] was held on 2009-10-27. The full log is available[13]. Edward Kirk asked if there was a firm date yet set for the semantics switchover (marking triaged bugs as NEW with the Triaged keyword rather than ASSIGNED). Adam Williamson looked at the schedule and noted it should be around 2009-11-12 if no further schedule changes occurred.

No-one had heard from Brennan Ashton regarding his promised summary of the status of the triage metrics project.

Edward Kirk wondered if the bug workflow page and diagram[14] would require updating when the semantics change occurred. Adam Williamson believed it would, but the necessary changes would be quite minor. Edward and Adam agreed to keep the necessary changes in mind for the meeting prior to the semantics change.

Edward Kirk promised to make sure the email warning developers that the regular housekeeping changes in Bugzilla at release time would be coming soon.

The next QA weekly meeting will be held on 2009-11-02 at 1600 UTC in #fedora-meeting, and the next Bugzappers weekly meeting on 2009-11-03 at 1500 UTC in #fedora-meeting. Note that the meeting times in UTC do not change even though many countries are going through daylight savings time changes around this time of year, with the result that the meetings will be one hour earlier for many people in practice.

  1. ↑ http://fedoraproject.org/wiki/QA/Meetings
  2. ↑ http://meetbot.fedoraproject.org/fedora-meeting/2009-10-26/fedora-meeting.2009-10-26-16.08.log.html
  3. ↑ http://fedoraproject.org/wiki/Category:Debugging
  4. ↑ http://fedoraproject.org/wiki/Test_Day:2009-10-22
  5. ↑ http://bugzilla.redhat.com/show_bug.cgi?id=528317
  6. ↑ http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=530452
  7. ↑ http://kparal.wordpress.com/2009/10/21/rpmguard-print-important-differences-between-rpms
  8. ↑ http://fedorahosted.org/autoqa/ticket/75
  9. ↑ http://wwoods.fedorapeople.org/files/critical-path/critpath.py
  10. ↑ http://fedorahosted.org/mailman/listinfo/autoqa-devel
  11. ↑ http://fedoraproject.org/wiki/Test_Day:2009-10-29
  12. ↑ http://fedoraproject.org/wiki/BugZappers/Meetings
  13. ↑ http://meetbot.fedoraproject.org/fedora-meeting/2009-10-27/fedora-meeting.2009-10-27-15.09.log.html
  14. ↑ http://fedoraproject.org/wiki/BugZappers/BugStatusWorkFlow

Fedora 12 testing

Much of the week's mailing list activity centred on testing the Fedora 12 Beta and post-beta updates, with much valuable testing being performed by many volunteers. Adam Williamson asked[1] for group members to provide feedback on the latest accepted kernel build, which had incorporated several changes from the kernel shipped in the Beta release. Many testers replied with helpful confirmation that the new kernel worked well. Liam Li announced[2] the pre-RC install testing cycle and associated test matrix[3], asking group members to try and cover as much of the install test case set as possible before the release candidate phase began on 2009-11-04; he later provided a report[4] on this testing. Adam requested testing[5] of an ext4 data corruption issue[6] which had surfaced in upstream kernel 2.6.32 testing to try and ensure that it was not affecting the 2.6.31 kernel included in Fedora 12.

  1. ↑ http://www.redhat.com/archives/fedora-test-list/2009-October/msg00650.html
  2. ↑ http://www.redhat.com/archives/fedora-test-list/2009-October/msg00721.html
  3. ↑ http://fedoraproject.org/wiki/Test_Results:Fedora_12_Pre-RC_Install
  4. ↑ http://www.redhat.com/archives/fedora-test-list/2009-October/msg00816.html
  5. ↑ http://www.redhat.com/archives/fedora-test-list/2009-October/msg00805.html
  6. ↑ http://bugzilla.kernel.org/show_bug.cgi?id=14354

Blocker bug review

The second Fedora 12 blocker bug review meeting took place on Friday 2009-11-30, and Adam Williamson posted a summary[1]. He noted that all remaining 43 blocker bugs had been reviewed, linked to the meeting summary[2] which outlined the status for each bug, and thanked the many members of the QA, release engineering and development groups who had contributed to the meeting.

  1. ↑ http://www.redhat.com/archives/fedora-test-list/2009-October/msg00845.html
  2. ↑ http://meetbot.fedoraproject.org/fedora-bugzappers/2009-10-30/fedora-bugzappers.2009-10-30-15.01.html

Fedora 10 bug review event

Edward Kirk announced[1] a BugZappers event on 2009-10-30 at which the group would gather to try and review remaining Fedora 10 bugs and see which could be either closed or promoted to Fedora 11 or 12, prior to the automated closing of these bugs as old when Fedora 12 is released.

  1. ↑ http://www.redhat.com/archives/fedora-test-list/2009-October/msg00791.html
Retrieved from "https://fedoraproject.org/w/index.php?title=FWN/Beats/QualityAssurance&oldid=133638"