Dresden Marathon 2013 (AOK 10km Lauf)

Letzten Sonntag war es wieder mal soweit. Der alljährliche Dresden Marathon inclusive Alternativstrecken (Halbmarathon und 10km-Lauf) fand statt. Auch ich war wieder mit dabei, musste mich aber jedoch schon vorab von meinem Ziel, einen Halbmarathon zu laufen, verabschieden. Aufgrund Zeitmangels in den letzten Monaten hatte ich nur ein einziges Mal die Chance, diese Distanz zu trainieren und hatte sie auch nur knapp verfehlt. Somit blieb für mich der 10km Lauf übrig. Immerhin die Möglichkeit meine Bestzeit vom letzten Jahr zu übertrumpfen?

Die Tage zuvor waren grau mit teilweise Regen. Düstere Gedanken machten sich breit, da ich nicht wirklich gern im Regen laufe, zumindest nicht zu dieser Jahreszeit mit ca. 10°C Außentemperatur. Aber wir hatten Glück und der Sonntag startete mit strahlendem Sonnenschein und wunderbaren Temperaturen. Aber auch die Masse an teilnehmenden Läufern war beträchtlich. Mehr als 9000 Laufbegeisterte jedes Alters fanden sich ein, um ein gemeinschaftliches Laufgefühl auszuleben. Es war einfach toll.

(c) http://www.dresden-marathon.com

Des Rennen an sich verlief ganz gut. Vor Allem aufgrund des großen Nachrückens im Startbereich, stand ich relativ weit vorn und die Menge an Läufern verteilte sich gut. Damit konnte ich schon hier mein Tempo von ca. 5:00 min/km durchziehen. Dies setzte sich dann gut über die nächsten Kilometer fort. Vorbei an der Semperoper und den großen Trauben anfeuernder Menschen, entlang des Terassenufers bis hinauf zum Sachsenplatz. Dort gab es wie jedes Jahr die Auftrennung zwischen (Halb-)Marathon und 10km. Ab hier wollte ich eigentlich das Tempo erhöhen, aber aufgrund des mangelnden Trainings der letzten Wochen, war ich nicht wirklich dazu in der Lage. Ein aufkommendes Seitenstechen setzte Riegel vor und ich fiel auf mein normales Tempo zurück. Besser so gut ankommen, als sich kilometerlang damit rumzuquälen. Ich war mächtig erstaunt, als ich kurz danach ein junges, zierliches Mädchen überholt habe, die vollen Eifers nicht viel langsamer als ich gerannt ist. Hut ab!

Weiter ging es nun über die Carolabrücke entlang des Rosengartens bis kurz vor die Albertbrücke, dort hinunter zum Elbufer und entlang des Elbradweges bis zum Palaisgarten. Beim letzten Überqueren der Elbe über die Augustusbrücke beschleunigte ich nun doch noch etwas, um wenigstens auf den letzten Kilometer noch eine bessere Zeit herauszuholen. Die jubelnden Massen im Verlauf der Zielgeraden machen es dann doch etwas einfacher, noch einmal Höchstleistung zu geben.

Letztendlich erreichte ich das Ziel in einer guten Verfassung mit 49:38min. Somit leider kein neuer Streckenrekord für mich selbst, aber doch eine Zeit, mit der ich zufrieden bin. Mit Platz 46 in meiner Altersgruppe mit 154 Läufern war dies somit im ersten Drittel und etwa gleichauf mit der Gesamtzahl männlicher Läufer (Platz 389 von 1328). Ich bin zufrieden.

Noch glücklicher bin ich, da ich zusammen mit meinen Spendern, aufgrund einer mit dem Lauf verbundenen Spendenaktion des PTV-Sachsen e.V. eine Spendensumme von 215 EUR erlaufen konnte. Die Summe kommt wieder Kindern von sozial schwachen Familien zu Gute. Also vielen, vielen Dank für eure Unterstützung.

Wer als Laufbegeisterter nun noch genaue Informationen zu meinem Lauf sehen möchte, der sollte bitte nachfolgend die Garmin-Daten anschauen:

Automation Development report – week 42 2013

Highlights

Henrik finally updated the current quarterly goals for our team. Please see our prioritized projects or check his blog post from today.

We were surprised to see a bug for the httpd.js component fixed after more than a year, which drastically improved the execution time of our Mozmill tests. Given that we are still not able to upgrade our CI system to Mozmill 2.0, another release off the 1.5 branch had to go out. So we made Mozmill 1.5.24 available for everyone.

With the help of IT, Henrik was able to make the new Windows 8.1 nodes available for our Mozmill-CI production system. That means we have 4 nodes for the 32bit version and another 4 nodes for 64bit. Sadly we are still facing bad “A critical MSR modification” system crashes for the 64bit nodes, where we think it might be related to our vSphere software, which doesn’t have full support for Windows 8.1 yet. If you have ideas please let us know on bug 916746.

Dave was working on b2gperf and got version 0.11 released.

Individual Updates

For more granular updates of each individual team member please visit our weekly team etherpad.

Meeting Details

If you are interested in further details and discussions you might also want to have a look at the meeting agenda and notes from this weeks Automation Development meeting.

Automation Development team goals for Q4 2013

I know that I’m a bit late in publishing our Automation Development team goals for the last quarter this year. But better late than never. I think there are some very interesting tasks we want to tackle on over the course of the next three months. And we would invite everyone who has interest to join us, learn new stuff, and simply have fun with coding.

Given that we are a part of the https://wiki.mozilla.org/Auto-tools team aka the A-Team, we inherit some of the overall team goals for Q4 2013. As of now all of our own work is concentrated around Desktop QA Automation and B2G (Firefox OS). Collecting those various pieces we came up with the following list of prioritized projects.

Cross-Team Goals

Given that one of our core principles is to take care of test automation across different teams, and to teach everyone with interests under this topic, we have some tasks which have to get finished off by working with people from a couple of other teams. Those are:

Switch Mozmill CI from Mozmill 1.5.x to Mozmill 2.0

After we have released Mozmill 2.0 by end of September we now have to work on getting our CI system upgraded too. The necessary work is listed on issue 329 of our issue tracker on Github.

There is not that much to do here beside fixing two bugs, which block us from an upgrade. That’s a recent change which comes with Firefox 26.0 and confused our restart test handling. For more details see bug 922995. And on the other side our update testrun leaves mounted DMG images behind on OS X 10.6. Therefore see issue 90.

Get FirefoxMetro tests executed in Mozmill CI production

Given that the QA team gave us a couple of Metro tests to automate, we have to add support to run all of them in our CI system. There is no support for that type of application yet. So a good amount of work lays in-front of us. All tasks are covered by issue 215.

Mozmill 2.0 already comes with MetroFirefox support but still misses some critical pieces. As of now it’s not able to set the installed application as the default browser on the system. Starting with Windows 8 this system level change is way harder to do. If anyone of you has ideas in how to do that please let us know.

AWSY (Are we slim yet) Firefox OS Memory Tests

We have to develop test automation to add Firefox OS to http://www.areweslimyet.com. The test, which is an orangutan-based script driving the gaia-ui, will be run on the B2G emulator, using TBPL builds from mozilla-central. Memory dumps will be grabbed and results posted to AWSY, with the ultimate goal in identifying and preventing Firefox OS memory leak regressions. For more information please see the tracking bug.

Get SoftVision to lead the automation efforts for QA with our oversight

Our team members from the test automation team at Softvision are helping us for a long time now in maintaining the Mozmill tests repository by fixing broken tests and creating new ones. Also they have assisted in the final work for the Mozmill 2.0 release, and other tools we own. To give them more responsibility we want to bring all of them into the position to handle the Mozmill-CI system on their own. This is necessary given the much broader scope of tasks everyone in our team is assigned to. Therefore we will do trainings and demos in our Ask an expert meetings.

Team Goals

There are also some additional goals, which do not scale across teams, but are also important to note.

Deploy Eideticker CI system for Firefox OS

We have to implement a continuous integration system for the Firefox OS start-up and scrolling tests using an external camera. More details about this goal can be found on bug 904837.

Extend Mozmill CI to cover most of our localized builds

To lower the amount of possibly broken beta and release builds we have to extend our tests to cover nearly all supported locales. Therefore we have to bring more nodes online in our CI system, and further enhance the tests to run more efficient. The ultimate goals would be to test all of our >90 locales. If you are interested to help us please let us know or check issue 270.

Create web application for on-demand update jobs

We want to reduce the failure rate when setting up the configuration for on-demand update tests in our Mozmill-CI system. Therefore a simple web application has to be created which allows us to pick the configuration like what can be done with the trychooser. If you want to help please check bug 915164

Stretched Goals

Last but not least we also have goals which are stretched and we might work on if there is time left.

Setup try-server like Mozmill CI system

To help everyone who is working on creating new or fixing broken Mozmill tests, we want to have a try server like system. That means you could push your patch and it will be tested across platforms and Firefox versions.

Investigate requirements for community driving tool

One of the goals from Henrik is to lower the amount of volunteers who are leaving our team or even Mozilla because of lack of activity from our side. Given that there is no good way in keeping in touch with everyone yet, a tool has to be created or extended. So maybe we can chime into the Black Hole.

As you can see there is a lot of work, which is awaiting to be finished by this quarter. We always appreciate your willingness and interest in helping us. So if one or some of those goals have drawn your interest let us know.

Automation Development report – week 41 2013

Similar to last week a couple members from our team were partly away last week given the return travels from the Mozilla Summit. So we have lesser announcements to make, except that we hopefully found a couple of new volunteers for test automation! Which would be fantastic.

Highlights

As of last week we got our new Mac Minis in SCL3 setup by IT. That means that we have expanded our Mozmill-CI to also run pulse triggered tests for all supported versions of Firefox on OS X 10.9. Since then 2 betas have been tested and no failures occurred. I call this a great start! Further I have prepared a further single Mac Mini for each 10.6, 10.7, and 10.8. So we have 4 nodes per OS X version to run tests on. Linux and Windows will follow shortly.

Further a fix has been landed for httpd.js, which solves one of our pain points in using that server. With it we can save a lot of time, and as already posted yesterday, we have released Mozmill 1.5.24 for it.

Further please welcome our new contributor Kamil Kjozwiak, who wants to help us with the creation of Mozmill tests for Firefox Metro.

Individual Updates

For more granular updates of each individual team member please visit our weekly team etherpad.

Meeting Details

If you are interested in further details and discussions you might also want to have a look at the meeting agenda and notes from this weeks Automation Development meeting.

Mozmill 1.5.24 released

As of yesterday we have released Mozmill 1.5.24. That release was a bit unexpected because we had already decided to not release another version of this branch. But we were able to include a bug fix, which let our tests run nearly twice that fast. And given that we cannot upgrade our Mozmill CI to version 2.0 yet, a quick 1.5 release provides us a huge performance win.

The problem in detail was in the local HTTP server we make use of, which is httpd.js. In July 2012 I recognized a drastic speed decrease, and figured out that this server takes about 6s to shutdown if the same locally served test page is used multiple times. Given that we still start and stop httpd.js in each of the tests, this sums up to a really annoying delay. As it has turned out it actually was a regression introduced by a change in Firefox and now after 15 months bug 777354 got finally fixed! So huge thanks goes to Patrick McManus here!

As usual you can get the new release via pypi or install it via ´pip install –upgrade mozmill==1.5.24´. For our mozmill-automation users a synonym package has also been released to pypi.

For more information about this Mozmill release please see bug 926909.

Automation Development report – week 40 2013

Given that most people of our team were out for the Mozilla Summit over the last weekend, we weren’t able to get a lot of tasks accomplished. But compared to the awesome time and more a dozen of discussions about automation with people during the event, we will most likely see some new faces soon.

Hightlights

As of Wednesday last week we updated our Mozmill CI instance to make use of Mozmill 2.0. Sadly we had to roll back this change, given that our restart tests were failing all over the place for Firefox Nightly and Aurora. An early investigation has been shown that a change in Firefox has caused this misbehavior of Mozmill. Hopefully we can get this fully investigated and fixed soon.

For Gaia related work Dave has released a couple of interesting software updates, which are gaiatest 0.16, b2gpopulate 0.10, and b2gperf 0.10.

Beside that you might also want to have a look at Dave’s blog post about the new command line interface tool for Gaia, which let you run commands directly via the command line.

Individual Updates

For more granular updates of each individual team member please visit our weekly team etherpad.

Automation Development report – week 39 2013

During the last week we did something phenomenal. As you might have heard, we were able to release Mozmill 2.0 to the world. After about 3 years of development it was one of our biggest milestones to finish off. But also lots of trouble with our Windows nodes in Mozmill-CI caused us a bit of headache, which we were able to finally get solved. For more details please read on…

Hightlights

As already announced by my blog post last week, we were able to finally release Mozmill 2.0. Since our team has taken over the work, we put a lot of additional efforts into the project. That said we really want to say a huge THANK YOU to all contributors, who helped us to finish off that work. You all are totally amazing! With the new version we will have a way more stable and feature-rich functional test framework available. Don’t miss to get it installed from pypi and try it out yourself. Keep in mind that it is fully backward compatible with Mozmill 1.5.x! So all of your existent tests will continue to work without modifications.

Beside that a lot of trouble came over us in terms of mozmill-ci breakage. As what I wrote in the last Automation Development report, we were facing a lot of issues with all of our Windows nodes. The high CPU load caused tests to run very slow and we even faced application disconnect issues with Firefox due to no response coming back to Mozmill. Henrik was heavily working on to get those issues fixed. The first step in disabling the Windows 8.1 nodes were already a good start, and we got more stable testruns. Now we are waiting for the VMs with the final version of Windows 8.1, which will hopefully be ready soon. But also disabling the Windows Defender helped a lot; we really don’t need that on our nodes for testing Firefox. So especially the suspicious installation and uninstallation duration for Firefox went down from about 4 minutes to only a couple of seconds. That’s the best time saver at all! Further we disabled the Windows Error Reporting, which came into the way because the huge amount of crashes we experienced lately. With that nothing halts the execution of our tests. A big thank you for the hint goes to Armen from the RelEng team!

Dave continued his work on Firefox OS and got some important new releases out for b2gperf 0.8, gaia-test 0.14 and gaia-test 0.15. Also you should have a look at this blog post for Building a harness for Eideticker… with LEGO which is kinda amazing. If you want to see it in real life and you are participating the Mozilla Summit in Brussels, check out his booth at the innovation fair.

Individual Updates

For more granular updates of each individual team member please visit our weekly team etherpad.

Double-check to disable Firefox Sync when doing a regression test to not loose important profile data

Yesterday I was doing a regression test for Firefox to determine the changeset which brought in a very annoying behavior into Aurora builds. Therefore I copied my profile to not destroy my original profile data, and was working on that copy. All was working fine and I was able to reduce the whole profile to only the sessionstore.js and prefs.js files.

The evil awakening came today morning when I tried to log into a website via my personal profile. By surprise the user name and password hasn’t been filled in automatically, and a check for the saved passwords have revealed that NONE are stored anymore! All were gone forever.

After some thoughts I figured out that sync was the problem here, because I missed to disconnect it in the copied profile. So after I have deleted the signons.sqlite and other files it also removed all the information from my sync account, which got distributed to all my other machines too. So on each of them the passwords are gone.

And if that isn’t bad enough, the backup tool on my Ubuntu machine had a hick-up lately and deleted all the old backup files on the backup drive. Not sure how that have come, but with that I completely lost all my stored passwords, and other synced data. :(

So a warning for everyone who is doing regression tests for Firefox and has Sync enabled: Please do NOT forget to disconnect Sync for the testing profile and double-check that it has been turned off.