Filed by: sstanca [at] mozilla.com
Parsed log: https://treeherder.mozilla.org/logviewer?job_id=409936309&repo=autoland
Full log: https://firefox-ci-tc.services.mozilla.com/api/queue/v1/task/EExl1SIxTfi4cE3rR008vg/runs/0/artifacts/public/logs/live_backing.log
Reftest URL: https://hg.mozilla.org/mozilla-central/raw-file/tip/layout/tools/reftest/reftest-analyzer.xhtml#logurl=https://firefox-ci-tc.services.mozilla.com/api/queue/v1/task/EExl1SIxTfi4cE3rR008vg/runs/0/artifacts/public/logs/live_backing.log&only_show_unexpected=1

[task 2023-03-23T14:35:36.123Z] 14:35:36     INFO - TEST-OK | dom/tests/mochitest/webcomponents/test_custom_element_ensure_custom_element.html | took 146ms
[task 2023-03-23T14:35:36.203Z] 14:35:36    ERROR - TEST-UNEXPECTED-FAIL | chrome://mochitests/content/chrome/dom/tests/mochitest/webcomponents/test_custom_element_ensure_custom_element.html | changed preference: media.gmp.storage.version.observed
[task 2023-03-23T14:35:36.204Z] 14:35:36     INFO - TEST-START | Shutdown

:tgiles, since you are the author of the regressor, bug 1813077, could you take a look?

For more information, please visit auto_nag documentation.

I have no idea how the newly added test_custom_element_ensure_custom_element is the culprit in this case. The test does not modify any preferences nor accesses the media.gmp.storage.version.observed preference (at least not directly). This test has nothing to do with the Gecko Media Plugin service. The newly added ensureCustomElements function does not modify (as far as I can see) the previously mentioned media pref either, no code I added has direct ties to the GMP (but I don't have experience with the GMP so I could be missing some connections).

I'm not sure what this test could be doing to cause GMP storage to have "an incompatible version, and if so nuke it". The only time we would change the version.observed value is when the observed version and the expected version are different. I'm guessing, since this is native code, the OS or other programs can do something to change this value, at least this is what I'm gleaning from Bug 1270968. I'm not sure why this storage version value would change in our CI environment.

Like looking at some of the other failures, it seems random prefs are changing their values and I'm not sure why that is happening...for example, here is a Windows run that shows "browser.bookmarks.restore_default_bookmarks" being changed. Hopefully I can find someone who understands what is going on here.

Depends on D173470

Not sure if we would call Bug 1783248 the regressor of this bug, but that change is what is causing this test-verify failure. Added the media.gmp.storage.version.observed to the ignorePrefs list, which seems like the solution for this bug.

Set release status flags based on info from the regressing bug 1813077

https://hg.mozilla.org/mozilla-central/rev/781b378ff938
https://hg.mozilla.org/mozilla-central/rev/3f0e496de1fe