The sources.xml that is generated by releng for unagi builds does not contain the gecko commit that was used to produce the build. We have some automation that currently consumes releases.m.c builds that uses this information; we can't switch away from those builds to official releng builds until sources.xml contain the gecko commit. In bug 820245, it seemed that this was added, but apparently only to unagi builds, and not unagi-eng builds. Compare: https://pvtbuilds.mozilla.org/pub/mozilla.org/b2g/nightly/mozilla-b2g18-unagi-eng/latest/sources.xml https://pvtbuilds.mozilla.org/pub/mozilla.org/b2g/nightly/mozilla-b2g18-unagi/latest/sources.xml Note that the gecko commit is in an XML comment, which may make it harder to parse in automation. Is there any reason not to make this a normal XML element?
I think we're using comments since sources.xml is for 'repo', which probably won't expect unknown xml elements.
Created attachment 709817 [details] [diff] [review] get unagi-eng manifest updated+uploaded I think this'll do the trick.
Assignee: nobody → aki
Attachment #709817 - Flags: review?(catlee)
Comment on attachment 709817 [details] [diff] [review] get unagi-eng manifest updated+uploaded http://hg.mozilla.org/build/mozharness/rev/9ca985b8d450 Merged to production.
Attachment #709817 - Flags: checked-in+
Kicked off nightlies on b2g18.
a) i typo'ed the fix, should s,update,upload, in the second action in the patch. b) we didn't actually translate hg to git. Did you need the hg gecko sha, or the git one?
Landed typo fix: http://hg.mozilla.org/build/mozharness/rev/1d94b75d4731
(In reply to Aki Sasaki [:aki] from comment #5) > a) i typo'ed the fix, should s,update,upload, in the second action in the > patch. > b) we didn't actually translate hg to git. Did you need the hg gecko sha, > or the git one? I think it doesn't particularly matter at this point; we just need to be able to distinguish different builds based on commit. We could use either git-based or hg-based commits for this.
Ok, these are uploading as "unagi", which is probably problematic. I need to upload these as "unagi-eng".
Created attachment 709949 [details] [diff] [review] fix unagi-eng manifest target_suffix per http://hg.mozilla.org/build/mozharness/file/0f9a9e51d309/scripts/b2g_build.py#l955
Attachment #709949 - Flags: review?(catlee)
Comment on attachment 709949 [details] [diff] [review] fix unagi-eng manifest target_suffix http://hg.mozilla.org/build/mozharness/rev/735d97a6938f
Attachment #709949 - Flags: checked-in+
Status: NEW → RESOLVED
Last Resolved: 5 years ago
Resolution: --- → FIXED
FWIW, the |"translate_hg_to_git": True,| line needs to be a child of "manifests" for the hg to git translation to be enabled. Currently it's not there, so we just get hg revisions and slightly misleading config.
(In reply to Nick Thomas [:nthomas] (gone 14-24 Feb PST) from comment #12) > FWIW, the |"translate_hg_to_git": True,| line needs to be a child of > "manifests" for the hg to git translation to be enabled. Currently it's not > there, so we just get hg revisions and slightly misleading config. Do we need to reopen this, then?
Per comment #7 it doesn't matter for the manifests, I mention it from a RelEng point of view after getting confused when diffing against another config.
Product: mozilla.org → Release Engineering
You need to log in before you can comment on or make changes to this bug.