All posts by Henrik Skupin

Firefox Automation report – week 39/40 2014

In this post you can find an overview about the work happened in the Firefox Automation team during week 39 and 40.

Highlights

One of our goals for last quarter was to get locale testing enabled in Mozmill-CI for each and every supported locale of Firefox beta and release builds. So Cosmin investigated the timing and other possible side-effects, which could happen when you test about 90 locales across all platforms! The biggest change we had to do was for the retention policy of logs from executed builds due to disk space issues. Here we not only delete the logs after a maximum amount of builds, but also after 3 full days now. That gives us enough time for investigation of test failures. Once that was done we were able to enable the remaining 60 locales. For details of all the changes necessary, you can have a look at the mozmill-ci pushlog.

During those two weeks Henrik spent his time on finalizing the Mozmill update tests to support the new signed builds on OS X. Once that was done he also released the new mozmill-automation 2.0.8.1 package.

Individual Updates

For more granular updates of each individual team member please visit our weekly team etherpad for week 39 and week 40.

Meeting Details

If you are interested in further details and discussions you might also want to have a look at the meeting agenda, the video recording, and notes from the Firefox Automation meetings of week 39 and week 40.

Firefox Automation report – week 37/38 2014

In this post you can find an overview about the work happened in the Firefox Automation team during week 37 and 38.

Highlights

After 7 months without a new release we finally were able to release mozdownload 1.12 with a couple of helpful enhancements and fixes.

We released Mozmill 2.0.7 and mozmill-automation 2.0.7 mainly for adding support of the v2 signed Firefox application bundles on OS X. Sadly we quickly had to follow-up with an appropriate 2.0.8 release for both tools, because a let change in the JS Engine caused a complete bustage of Mozmill. More details can be found in my appropriate blog post.

We were even able to finally release Memchaser 0.6, which is fixing a couple of outstanding bugs and brought in the first changes to fully support Australis.

One of our goals was to get the failure rate of Mozmill tests for release and beta candidate builds under 5%. To calculate that Cosmin wrote a little script, which pulls the test report data for a specific build from out dashboard and outputs the failure rate per executed testrun. We were totally happy to see that the failure rate for all Mozmill tests was around 0.027%!

During the merge process for the Firefox 32 release Andrei has seen some test inconsistencies between our named branches in the Mozmill-Tests repository. Some changes were never backported, and only present on the default branch for a very long time. He fixed that and also updated our documentation for branch merges

Something else worth for highlighting is also bug 1046645. Here our Mozmill tests found a case when Firefox does not correctly show the SSL status of a website if you navigate quickly enough. The fix for this regression caused by about:newtab made it even into the release notes

Last but not least Andreea started planning our Automation Training day for Q3. So she wrote a blog post about this event on QMO.

Individual Updates

For more granular updates of each individual team member please visit our weekly team etherpad for week 37 and week 38.

Meeting Details

If you are interested in further details and discussions you might also want to have a look at the meeting agenda, the video recording, and notes from the Firefox Automation meetings of week 37 and week 38.

Firefox Automation report – week 35/36 2014

In this post you can find an overview about the work happened in the Firefox Automation team during week 35 and 36.

Highlights

Due to a lot of Mozmill test failures related to add-on installation and search, we moved from the addons-dev.allizom.org website to the staging website located at addons.allizom.org. Since then we experiencing way lesser test failures, which are most likely network related.

In order to keep up with all test failures, and requests for new tests we started our Bug triage meeting on a weekly basis on Friday. For details we have created an etherpad.

If you are interested in helping us with Mozmill tests, you can now find a list of mentored and good first bugs on the bugsahoy web page.

Because of the app bundle changes on OS X, which were necessary due to the v2 signing requirements, we had to fix and enhance a couple of our automation tools. Henrik updated mozversion, mozmill, and mozmill-automation. We were close to releasing Mozmill 2.0.7.

Individual Updates

For more granular updates of each individual team member please visit our weekly team etherpad for week 35 and week 36.

Meeting Details

If you are interested in further details and discussions you might also want to have a look at the meeting agenda, the video recording, and notes from the Firefox Automation meetings of week 35 and week 36.

Erster offizieller Halbmarathon

Nachdem ich mir bereits vor mehr als einem Jahr vorgenommen hatte, einen Halbmarathon zu laufen, hat es nun doch bis letzte Woche gedauert. Eine wirklich lange Zeit, aber der Grund waren einfach Knieprobleme, die mich letztes Jahr zu einer mehrmonatigen Pause beim Laufen zwangen. Aufgeben wollte ich trotzdem nicht und habe deshalb die Auszeit genutzt, und meine Füße im normalen Alltag komplett an das Barfußlaufen zu gewöhnen. Für jedmögliche Gelegenheit gibt es nun spezielle, sogenannte Barfußschuhe in meinem Schuhschrank. Durch deren ständiges Tragen konnte ich die Muskulatur in meinen Füßen wunderbar stärken und auch den Laufstil anpassen. Vom vormals Hinterfußläufer bin ich nun zum Mittelfuß- aber auch schon zum Teil Vorderfußläufer geworden. Diese Veränderung hat durch die natürliche Dämpfung des Fußes wesentlich geringere Belastungen für das Knie zur Folge, und lässt mich erfolgreich mit dem Training fortsetzen. Wie froh ich darüber bin, muss ich sicherlich nicht gesondert erwähnen. Seit ca. April dieses Jahres bin ich nun wieder wöchentlich aktiv und habe die gelaufenen Distanzen mit den Barfußschuhen kontinuierlich erweitert. Am 05.10.2014 hatte ich dann meine erste Halbmarathondistanz von 21.1km erfolgreich absolviert und war mir sicher, den offiziellen Lauf auch bewältigen zu können.

Starter des Marathons und Halbmarathons – (c) Bild von mopo24.de

Der 16. Dresdner Morgenpost Marathon fand am 19.10.2014 hier in Dresden statt. Dieses Mal mit einer komplett geänderten Streckenführung und einem getrennten Start der Marathon- und Halbmarathonläufer vom 10km Feld. Das war sicherlich hilfreich, aber aufgrund der riesigen Anzahl von Läufern – allein 2255 für den Halbmarathon – gab es trotzdem ein arges Gedränge am Start. Glücklicherweise konnte die breite Ostra-Allee und der weitere Weg über den Theaterplatz und die Augustusbrücke, die Läufermasse etwas auseinander ziehen, sodass spätestens in der Neustadt ein angenehmes Laufen möglich war. Auch das Wetter spielte wieder mit, das mit ca. 20 Grad und viel Sonne ein eher sommerliches Ambiente schuf.

Dieses Mal musste ich nicht allein rennen, sondern hatte Anita als Begleitung dabei. Für sie war es bereits der zweite Halbmarathon. Ich wollte den Lauf unterhalb von 2h beenden, und von ihrer Seite war ein Pace von 5:30 min/km geplant. Da beide nicht weit auseinander liegen, schloss ich mich ihr also an. Anita musste mich aber gerade zu Beginn gut drosseln. Wie immer wollte mich die schnellere Masse mitziehen. Nach den ersten Kilometern waren wir aber gut in unserem Tempo drin, und ich habe dann auch öfters selbst mal die Zeit kontrolliert. Alles lief ziemlich gut, bis bei Kilometer 13 dann die ersten Probleme mit dem linken Fuß begannen. Das ständige Laufen auf Asphalt ist mit Barfußschuhen dann doch etwas anderes, als mein Training auf gemischtem Boden. Aber es half nichts. Mein Wille war da, es zu schaffen! Deshalb wurde mit zusammengebissenen Zähnen weitergerannt. Wenn ich die geplante Zeit nicht schaffe, dann wollte ich wenigstens das Ziel erreichen. Glücklicherweise ließen die Schmerzen auf den letzten beiden Kilometern etwas nach. Ein Abgleich mit der Zeit bei Kilometer 20 zeigte aber auch, dass wir schon lange nicht mehr den 5:30 min/km Pace hatten. Um ihn doch noch zu schaffen, nahm ich bei etwas mehr als 500m vorm Ziel all meine Kräfte zusammen und sprintete ins Ziel. Das Resultat war unglaublich – mit 5:30 min/km Pace war ich somit exakt nach 1:56:50 anggekommen. Ein einfach herrliches Gefühl sich selbst noch etwas mehr als die 2h unterboten zu haben.

Selbstportrait kurz vor dem Endspurt – (c) V. Lichanov

Ganz besonders freut es mich natürlich wieder, dass die mit dem Lauf verbundene Spendenaktion zusammen mit dem PTV Sachsen so erfolgreich war. Aufgrund all der engagierten Spender, die meinen Lauf unterstützt haben, kamen stolze 219 EUR für einen wohltätigen Zweck zusammen. Sicherlich wird ein Großteil wieder an Aktionen für Kinder aus sozial schwächeren Familien genutzt werden. Somit vielen lieben Dank an Euch alle!!!

Nach diesem Lauf und dem Erreichen einer neuen Distanz und Zeit, liegen nun neue Herausforderungen vor mir. Sicherlich werde ich jetzt versuchen, Zeiten für die 10km und Halbmarathon-Distanz zu verbessern, aber ein kompletter Marathon-Lauf wäre auch ein klasse Ziel – vielleicht ja schon für das nächste Jahr zur gleichen Zeit!

Firefox Automation report – week 33/34 2014

In this post you can find an overview about the work happened in the Firefox Automation team during week 33 and 34.

Highlights

To make sure that our weekly meetings will be more visible to our community, we got them added to the community calendar. If you are interested in what’s going on for Firefox Automation you are welcome to join our Monday’s team meeting.

In regards of the Mozmill project, Henrik landed his patch, which makes Mozmill more descriptive in terms of unexpected application shutdowns. Especially in the past weeks we have seen that Firefox does not restart as expected, but simply quits. There is bug 1057246 filed for the underlying problem. So with the patch landed, Mozmill will log that correctly in the results. Beside that we can also better see when crashes or a not by Mozmill triggered quit happens.

For Mozmill CI we landed a couple of enhancements and fixes. The most important ones were indeed the addition of 20 new locales for testing beta and release builds of Firefox across supported platforms. That means we cover 30 of about 95 active locales now. To cover them all, a good amount of follow-up work is still necessary. Immediately we stopped to run add-on tests for all branches except Nightly builds to save more time on our machines.

Henrik also continued on PuppetAgain integration for our staging and production CI systems. One of the blockers was the missing proxy support, but with the landing of the patch on bug 1050268 all proxy related work should have been done now.

Also on the continuous integration for TPS tests we made progress. The implementation got that far for Coversheet that we made the Jenkins branch the active master. There are still issues to implement or get fixed before the Jenkins driven CI can replace the old hand-made one.

Individual Updates

For more granular updates of each individual team member please visit our weekly team etherpad for week 33 and week 34.

Meeting Details

If you are interested in further details and discussions you might also want to have a look at the meeting agenda, the video recording, and notes from the Firefox Automation meetings of week 33 and week 34.

Firefox Automation report – week 31/32 2014

In this post you can find an overview about the work happened in the Firefox Automation team during week 31 and 32. It’s a bit lesser than usual, mainly because many of us were on vacation.

Highlights

The biggest improvement as came in during week 32 were the fixes for the TPS tests. Cosmin spent a bit of time on investigating the remaining underlying issues, and got them fixed. Since then we have a constant green testrun, which is fantastic.

While development for the new TPS continuous integration system continued, we were blocked for a couple of days by the outage of restmail.net due to a domain move. After the DNS entries got fixed, everything was working fine again for Jenkins and Mozilla Pulse based TPS tests.

For Mozmill CI we agreed on that the Endurance tests we run across all branches are not that useful, but only take a lot of time to execute – about 2h per testrun! The most impact also regarding of new features landed will be for Nightly. So Henrik came up with a patch to only let those tests run for en-US Nightly builds.

Individual Updates

For more granular updates of each individual team member please visit our weekly team etherpad for week 31 and week 32.

Meeting Details

If you are interested in further details and discussions you might also want to have a look at the meeting agenda, the video recording, and notes from the Firefox Automation meetings of week 32. There was no meeting in week 31.

Memchaser 0.6 has been released

The Firefox Automation team would like to announce the release of memchaser 0.6. After nearly a year of no real feature updates, but also some weeks of not being able to run Memchaser in Firefox Aurora (34.0a2) at all (due to a regression), we decided to release the current state of development as a public release. We are aware that we still do not fully support the default Australis theme since Firefox 29.0, but that’s an issue, which takes some more time to finish up.

Changes in 0.6

  • Upgrade to Add-on SDK 1.17 (#201)
  • Fix test_start_stop_logging for ‘File not found’ error (#199)
  • contentURL expects a string in Panel() and Widget() constructors (#198)
  • Bring the minimizeMemory() implementation up to date (#193)
  • Use ‘residentFast’ instead of ‘resident’ for memory reporter (#192)
  • Use postMessage for widget and panel communication instead port object (#185)
  • Support incremental cycle collection statistics (#187)
  • Added a usage section to the README (#178)
  • Change require statements and update the SDK to 1.14 (#174)
  • Simplify .travis.yml (#172)
  • Use . (dot) to include files while invoking /bin/sh (#169)
  • Use failonerror attribute in exec tasks (#168)

For all the details about the 0.6 release please check our issue tracker on Github.

mozdownload 1.12 has been released

The Firefox Automation team would like to announce the release of mozdownload 1.12. Without any other release of our universal download tool in the last 7 months, a couple of nice new features and bug fixes will make this release even more useful. You can upgrade your installation easily via pip, or by downloading it from PyPI.

Changes in 1.12

  • Display selected build when downloading (#149)
  • Add support for downloading B2G desktop builds (#104)
  • Download candidate builds from candidates/ and not nightly/ (#218)
  • Add Travis CI build status and PyPI version badges to README (#220)
  • Add Python 2.6 to test matrix (#210)
  • Fix broken download of mac64 tinderbox builds (#144)
  • Allow download even if content-length header is missing (#194)
  • Convert run_tests script to Python (#168)
  • Ensure that –date option is a valid date (#196)

Mozmill 2.0.7 and 2.0.8 have been released

The Firefox Automation team would like to announce the release of Mozmill 2.0.7 and Mozmill 2.0.8. Both versions had to be released in such a short time frame to ensure continuing support for Firefox. Some latest changes done for Firefox Nightly broke Mozmill, or at least made it misbehaving. If you run tests with Mozmill ensure to upgrade to the latest version. You can do this via PyPI, or simply download the already pre-configured environment.

Changes in 2.0.7

  • Bug 1066295 – testMultipleLoads.js is failing due to new HTTP Cache v2
  • Bug 1000098 – Fix testPageLoad.js test for invalid cert page
  • Bug 1065436 – Disable e10s until full support landed
  • Bug 1062773 – Disconnect errors invalidate the report
  • Bug 999393 – Expose assert and expect by default in sub modules
  • Bug 970820 – Mozmill aborts with socket.timeout when trying to send the report

Changes in 2.0.8

  • Bug 1067939 – JSBridge and Mozmill broken due to ‘let’ changes in bug 1001090

Please keep in mind that Mozmill 2.0 does not support electrolysis (e10s) builds of Firefox yet. We are working hard to get full support for e10s added, and hope it will be done until the next version bump mid of October.

Thanks everyone who was helping with those releases!

Firefox Automation report – week 29/30 2014

In this post you can find an overview about the work happened in the Firefox Automation team during week 29 and 30.

Highlights

During week 29 it was time again to merge the mozmill-tests branches to support the upcoming release of Firefox 31.0. All necessary work has been handled on bug 1036881, which also included the creation of the new esr31 branch. Accordingly we also had to update our mozmill-ci system, and got the support landed on production.

The RelEng team asked us if we could help in setting up Mozmill update tests for testing the new update server aka Balrog. Henrik investigated the necessary tasks, and implemented the override-update-url feature in our tests and the mozmill-automation update script. Finally he was able to release mozmill-automation 2.6.0.2 two hours before heading out for 2 weeks of vacation. That means Mozmill CI could be used to test updates for the new update server.

Individual Updates

For more granular updates of each individual team member please visit our weekly team etherpad for week 29 and week 30.

Meeting Details

If you are interested in further details and discussions you might also want to have a look at the meeting agenda, the video recording, and notes from the Firefox Automation meetings of week 29 and week 30.