From Fedora Project Wiki

< FWN‎ | Beats

Revision as of 05:23, 6 February 2010 by Adamwill (talk | contribs) (qa beat for 212)

QualityAssurance

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

Contributing Writer: Adam Williamson

Test Days

Last week's Test Day was on the introduction of NFSv4 by default in Fedora 13[1]. The NFS maintainer, User:stevedSteve Dickson, was kind enough to provide several automated test suites, and a good turnout of testers ran them on a variety of NFS configurations, providing valuable results.

No Test Day is currently planned for this 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[2].

Weekly meetings

The QA group weekly meeting[1] was held on 2010-02-01. The full logs are available[2]. James Laska reported that Christopher Beland had added links to the Bugzilla common_bugs queries to the Fedora 13 common bugs page[3].

Kevin Fenzi promised to look at updating the QA IRC bot, zodbot, to monitor the Fedora 13 blocker bugs instead of the Fedora 12 ones.

Will Woods and Kamil Paral reported that they had not yet discussed design ideas for the proposed AutoQA results database. Kamil felt that more people should be involved as the project would become important to several groups, if implemented. James Laska suggested a micro-FAD[4]. Kamil mentioned that Luke Macken had suggested looking at his Kobo project[5] for inspiration. Jesse Keating felt it might be over-engineered for the purposes of AutoQA. Will and Kamil agreed to organize a meeting during the week to begin designing the system.

Adam Williamson reported that he had moved the privilege escalation policy discussion to the development mailing list[6], and would go through the same feedback/revision cycle there as he had on the test list before finally escalating the draft policy to FESco.

Adam Williamson reported on his and Rui He's progres with documenting installation testing as a QA group activity. They had finalized the draft installation validation testing page[7] and created a draft desktop validation testing page[8]. They had added a paragraph to the Join page[9] which briefly explains the testing and link to the two more detailed pages. James Laska asked if the QA schedule should be altered to refer to 'release validation' rather than 'installation validation', to leave room for non-installation testing, and Adam said he thought this would be a good idea.

James Laska reported that another rawhide acceptance testing event had come around during the previous week. This involves the creation and automated testing of a tree complete with installer. This time the acceptance test suite passed, but the installed system was unbootable due to a bug preventing the creation of the initramfs[10]. Jesse Keating pointed out that this meant the installer was ready for 'last known good' status, but not the package set, and further asked whether the 'last known good' idea is intended to refer only to the installer or also to an associated package tree. There was general agreement that the 'last known good' page should list the tested installer and package tree, and note that the good installer may work with different package trees but could not be guaranteed to. James and Adam Williamson agreed to work on the 'last known good' Wiki page with information provided by Jesse.

Will Woods and Kamil Paral provided an AutoQA update. Will had created a working prototype of the dependency checking test, based on yum, which was 147 lines of code and took around 20 seconds to run. James Laska mentioned that he was hoping to see the rpmfluff tool for generating fake test packages become an official Fedora package soon. Kamil went over some updates to rpmguard; he had made it notice when two packages it is asked to compare are identical, and compare a package only to the previous package from stable or updates (not updates-testing).

James Laska reported that Liam Li had added a dvd_install.py script to the autoqa repo[11]. He encouraged others to follow up with their thoughts on the script.

James Laska had continued working on the packaging plan for gwt[12] and hoped to be able to start packaging soon.

Kamil Paral mentioned that he was working on a package update acceptance test plan, and asked the group to provide any information they had on how other projects have approached this issue. Adam Williamson mentioned the Mandriva policies[13] [14] in the area, and suggested Kamil might talk to Vincent Danen, who had managed the update process for Mandriva before joining Red Hat.

The Bugzappers group weekly meeting[15] was held on 2010-02-02. The full log is available[16]. Adam Williamson admitted he had done nothing on the subject of bugs filed against orphaned packages.

Sergey Rudchenko told the group about a script he had written to clean up abrt backtraces, providing bug 558883[17] as an example. He was interested in extending the script to compute the similarity of any two given backtraces, and also store multiple backtraces offline for comparison. Adam Williamson and Matej Cepl suggested he talk to the abrt team about integrating his ideas into abrt itself.

The next QA weekly meeting will be held on 2010-02-08 at 1600 UTC in #fedora-meeting. The next Bugzappers weekly meeting will be held on 2010-02-09 at 1500 UTC in #fedora-meeting.

Nouveau 3D acceleration testing

Bruno Wolff noted[1] that experimental 3D acceleration for nouveau was now available in the mesa-dri-drivers-experimental package, and asked whether it was yet at a point where the developers would be interested in bug reports. Adam Williamson promised to pass the question along to the nouveau maintainer.

Fedora 13 Alpha blocker bug review meeting

Adam Williamson announced[1] and later recapped[2] the first blocker bug review meeting for Fedora 13. The summary of the meeting is available from meetbot[3]. All current alpha blocker bugs were reviewed at the meeting.