Jenkins is now blowing up for reasons unknown to me.
Desktop QUnit tests are failing all the time due to timeout reasons (I can sometimes replicate this locally - this might have been due to some core change?)
Mobile tests are sometimes passing and sometimes throwing errors due to no tests being run. See this commit for an example when the mobile tests passed but core changes failed: https://gerrit.wikimedia.org/r/#/c/92000/
and here where both sets of test failed: https://gerrit.wikimedia.org/r/#/c/92994/
This all started since the seemingly harmless https://gerrit.wikimedia.org/r/92804 got merged.
any ideas what is going on? This currently stops us merging any code (since we disabled the verified override)
Le 01/11/13 00:43, Jon Robson a écrit :
Jenkins is now blowing up for reasons unknown to me.
Desktop QUnit tests are failing all the time due to timeout reasons (I can sometimes replicate this locally - this might have been due to some core change?)
Mobile tests are sometimes passing and sometimes throwing errors due to no tests being run. See this commit for an example when the mobile tests passed but core changes failed: https://gerrit.wikimedia.org/r/#/c/92000/
and here where both sets of test failed: https://gerrit.wikimedia.org/r/#/c/92994/
This all started since the seemingly harmless https://gerrit.wikimedia.org/r/92804 got merged.
any ideas what is going on? This currently stops us merging any code (since we disabled the verified override)
The issue is a race condition between both MobileFrontend qunit jobs. The first job to complete would delete the symlink in Apache document root, causing the second job to fail.
I changed the "unique" id with:
https://gerrit.wikimedia.org/r/#/c/93030/
Retriggered the failing change https://gerrit.wikimedia.org/r/#/c/92000/ and it is passing now :-]
Sorry!
Thanks Antoine!! Yeyyy! Looks like they are all passing now as well so we should make the QUnit tests voting. On 1 Nov 2013 02:10, "Antoine Musso" amusso@wikimedia.org wrote:
Le 01/11/13 00:43, Jon Robson a écrit :
Jenkins is now blowing up for reasons unknown to me.
Desktop QUnit tests are failing all the time due to timeout reasons (I can sometimes replicate this locally - this might have been due to some core change?)
Mobile tests are sometimes passing and sometimes throwing errors due to no tests being run. See this commit for an example when the mobile tests passed but core changes failed: https://gerrit.wikimedia.org/r/#/c/92000/
and here where both sets of test failed: https://gerrit.wikimedia.org/r/#/c/92994/
This all started since the seemingly harmless https://gerrit.wikimedia.org/r/92804 got merged.
any ideas what is going on? This currently stops us merging any code (since we disabled the verified override)
The issue is a race condition between both MobileFrontend qunit jobs. The first job to complete would delete the symlink in Apache document root, causing the second job to fail.
I changed the "unique" id with:
https://gerrit.wikimedia.org/r/#/c/93030/
Retriggered the failing change https://gerrit.wikimedia.org/r/#/c/92000/ and it is passing now :-]
Sorry!
-- Antoine "hashar" Musso Mob.: +33 6 98 81 18 38 -- Skype: hashar
... And Antoine is a star :) Thanks so much for sorting this out and all the work you have done on Jenkins over the past year! On 1 Nov 2013 07:12, "Antoine Musso" amusso@wikimedia.org wrote:
Le 01/11/13 15:08, Jon Robson a écrit :
Thanks Antoine!! Yeyyy! Looks like they are all passing now as well so we should make the QUnit tests voting.
mwext-MobileFrontend-qunit-mobile is now voting :-]
Well done!
-- Antoine "hashar" Musso Mob.: +33 6 98 81 18 38 -- Skype: hashar
Mobile-l mailing list Mobile-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/mobile-l