Closed Bug 1201383 Opened 5 years ago Closed 5 years ago
Android l10n jobs fail to push new release to test channels
20:13:07 INFO - Calling Balrog release pusher script ... 20:13:07 INFO - Running command: ['python', '/builds/slave/rel-m-rel-and-api-9_rpk_1-0000/tools/scripts/updates/balrog-release-pusher.py', '--build-properties', '/builds/slave/rel-m-rel-and-api-9_rpk_1-0000/balrog_props.json', '--buildbot-configs', 'https://hg.mozilla.org/build/buildbot-configs', '--release-config', '/builds/slave/rel-m-rel-and-api-9_rpk_1-0000/build/buildbot-configs/mozilla/staging_release-fennec-mozilla-release.py', '--credentials-file', '/builds/slave/rel-m-rel-and-api-9_rpk_1-0000/oauth.txt', '--api-root', 'https://aus4-admin-dev.allizom.org/api', '--username', 'stage-ffxbld'] in /builds/slave/rel-m-rel-and-api-9_rpk_1-0000/build .... 20:13:07 INFO - Required option release_channel not present 20:13:07 ERROR - Return code: 1
Not sure why we do this in every repack, rather than in a updates builder, perhaps some confusion in bug 1019724. Anyway, the point of this bug is to remove the 20 mins or so we spend retrying.
Summary: Android l10n jobs failing to submit into balrog → Android l10n jobs failing to push new release
Working on a patch in staging.
Assignee: nobody → nthomas
I have a patch which passes the required information. The current blockage is adding archive.mozilla.org to the list of allowed domains, but I haven't gone ahead with that because that's not the right domain for staging. I'm not sure where it's coming from, and have bigger fish to fry at the moment.
Summary: Android l10n jobs failing to push new release → Android l10n jobs fail to push new release to test channels
The patch is on a code path used by android single-locale releases only, currently broken and there is no screaming. In staging this gets us as far as trying to POST into Balrog, but then it fails on archive.m.o not being in the whitelist (bug 1211297 should resolve that). If you're curious see http://dev-master2.bb.releng.use1.mozilla.com:8710/builders/release-mozilla-release-android-api-11_repack_1%2F1/builds/5/steps/run_script/logs/stdio FTR, I was wrong in comment #0 about doing this for every locale. mobile_l10n.py is submitting data for each of the locale, then runs the balrog pusher script once to do the top-level stuff in the blog (fileUrls etc).
Comment on attachment 8670677 [details] [diff] [review] [gecko] Pass --release-channel Pls see comment #4 for explanation.
Attachment #8670677 - Flags: review?(rail)
(In reply to Nick Thomas [:nthomas] from comment #4) > FTR, I was wrong in comment #0 about doing this for every locale. > mobile_l10n.py is submitting data for each of the locale, then runs the > balrog pusher script once to do the top-level stuff in the blog (fileUrls > etc). Oh silly me, confused myself. Once per repack we call balrog-release-pusher.py, and there are 20 repack jobs. cf Firefox which calls it once in the updates job. In other news, still hitting this in staging 20:41:50 INFO - Caught HTTPError: Couldn't update release: Release blob contains forbidden domain. because it's trying to use upload.ffxbld.productdelivery.stage.mozaws.net in fileUrls. Turns out this is just a staging problem. The copy of the script being called is in <jobdir>/tools, which is the production repo. Meanwhile in <jobdir>/build/tools there is my repo conatining the fix to use ftpServer (bug 1206041).
Comment on attachment 8670677 [details] [diff] [review] [gecko] Pass --release-channel https://hg.mozilla.org/integration/mozilla-inbound/rev/024f11388f99 https://hg.mozilla.org/releases/mozilla-aurora/rev/32a0cca876cb https://hg.mozilla.org/releases/mozilla-beta/rev/e7c32a898f4d https://hg.mozilla.org/releases/mozilla-release/rev/d57792e7b540
Attachment #8670677 - Flags: checked-in+
This can be verified in the next Fennec beta/release.
Working for production Balrog, then it fails to submit to staging. --> bug 1216381
Status: NEW → RESOLVED
Closed: 5 years ago
Resolution: --- → FIXED
Removing leave-open keyword from resolved bugs, per :sylvestre.
You need to log in before you can comment on or make changes to this bug.