Reftest logs include new logspam between each test: `console.error: Translations: (new Error("Unable to retrieve the translation models.", "resource://gre/actors/TranslationsParent.sys.mjs", 887))` and another line for Utils.sys.mjs
Categories
(Firefox :: Translations, defect)
Tracking
()
Tracking | Status | |
---|---|---|
firefox-esr102 | --- | unaffected |
firefox113 | --- | unaffected |
firefox114 | --- | unaffected |
firefox115 | --- | fixed |
People
(Reporter: dholbert, Assigned: gregtatum)
References
(Regression)
Details
(Keywords: regression)
Attachments
(1 file)
STR:
- Run some directory of reftests, e.g. here's a relatively short directory (6 tests):
./mach reftest layout/reftests/text-stroke/
- Look at your terminal output.
ACTUAL RESULTS:
After each REFTEST TEST-LOAD
line, the are these two lines:
console.error: (new Error("Unexpected content-type \"text/plain;charset=US-ASCII\"", "resource://services-settings/Utils.sys.mjs", 399))
console.error: Translations: (new Error("Unable to retrieve the translation models.", "resource://gre/actors/TranslationsParent.sys.mjs", 887))
I I'm pretty sure the second line (for TranslationsParent.sys.mjs) was recently-introduced. Not sure about the first one (for Utils.sys.mjs), & whether that's introduced at the same time / for the same reason or not.
EXPECTED RESULTS:
No such log-spam.
Filing in Firefox|Translation to start out, given the TranslationsParent.sys.mjs
filename.
Reporter | ||
Comment 1•1 year ago
|
||
This shows up in TreeHerder reftest logs, too -- e.g. this recent one, which has 2622 instances of the message Unable to retrieve the translation models
:
https://treeherder.mozilla.org/logviewer?job_id=417740634&repo=mozilla-central
https://firefoxci.taskcluster-artifacts.net/LOO7RqXJQ_mbNSkgu9kTwg/0/public/logs/live_backing.log
Reporter | ||
Comment 2•1 year ago
|
||
For comparion/bisection, this reftest log from April 25 has no such logging:
https://treeherder.mozilla.org/logviewer?job_id=413689151&repo=mozilla-central
https://firefoxci.taskcluster-artifacts.net/cEQaf--jSuW8raUBdFwjOg/0/public/logs/live_backing.log
It's just got back-to-back REFTEST TEST-LOAD
log lines, no console.error
interspersed between them.
I'll see if I can bisect...
Reporter | ||
Comment 3•1 year ago
|
||
Last good push on m-c:
https://treeherder.mozilla.org/jobs?repo=mozilla-central&searchStr=linux%2Cdebug%2Creftest&revision=c10ba3b6a8ecfaa3bebc0a33d8740e61594766c2
- Its linux debug R1 log: https://firefoxci.taskcluster-artifacts.net/bY2iHMOPTiW0qRWrv6xfQQ/0/public/logs/live_backing.log
- No console.error(...) after each TEST-LOAD line.
First bad push on m-c:
https://treeherder.mozilla.org/jobs?repo=mozilla-central&searchStr=linux%2Cdebug%2Creftest&revision=5eb0c339e4a6c919642f1e70d15347b766ceec00
- Its linux debug R1 log: https://firefoxci.taskcluster-artifacts.net/fyaCOOgYTKCiXxTXvJOh8g/0/public/logs/live_backing.log
- Both of comment 0's console.error(...) lines are logged after each TEST-LOAD line.
Reporter | ||
Comment 4•1 year ago
|
||
In that first-bad-push, this would've been from the stack of patches that landed in bug 1831963.
nordzilla or gtatum, could you take a look? This is adding a lot of noise to reftest logs.
Reporter | ||
Updated•1 year ago
|
Reporter | ||
Updated•1 year ago
|
Comment 5•1 year ago
|
||
Set release status flags based on info from the regressing bug 1831963
Reporter | ||
Comment 6•1 year ago
|
||
FWIW, I confirmed that the logspam does go away if I turn off the translations pref (e.g. doing s/true/false/
on the relevant line of all.js
).
Reporter | ||
Comment 7•1 year ago
•
|
||
Also, this affects other testsuites as well (not just traditional reftests).
It affects Web-platform-tests, e.,g. this log:
https://firefoxci.taskcluster-artifacts.net/Bukyir2lRNKIY7E5UNS3cg/0/public/logs/live_backing.log
And mochitests, e.g. this log:
https://firefoxci.taskcluster-artifacts.net/W4iN_Q8MQ6-3m4M4Uw8obg/0/public/logs/live_backing.log
And xpcshell tests, e.g. this log: [EDIT, my mistake -- xpcshell tests don't seem to be affected]
Reporter | ||
Updated•1 year ago
|
Comment 8•1 year ago
•
|
||
Thank you Daniel!
I first became aware of this yesterday regarding this bug:
https://bugzilla.mozilla.org/show_bug.cgi?id=1836183#c4
The issue is that we are reporting an error when RemoteSettings
is not available/fails and this noise appeared when we enabled Translations for Nightly and Early Beta recently in Bug 1831963 because RemoteSettings
is not available in tests.
I believe Greg is already working on a patch that will quiet this.
Reporter | ||
Comment 9•1 year ago
|
||
Thanks, Eric! Possibly related to bug 1834155 then (also about RemoteSettings
being unavailable and causing logspam).
Assignee | ||
Comment 10•1 year ago
|
||
This is a quick fix until we can address the underlying issue.
Updated•1 year ago
|
Assignee | ||
Comment 11•1 year ago
|
||
Our code controls the translations log message, but the Unexpected content-type
error is coming from RemoteSettings. Somehow the network response is mocked in tests with a blank response, and it's being reported as an error. I think Bug 1836505 will stop that one from being reported as well.
Reporter | ||
Comment 12•1 year ago
|
||
Thanks!
Assignee | ||
Comment 13•1 year ago
|
||
I'm landing this one since it's not risky, but I confirmed that Bug 1836505 fixes the rest of the stray output. I plan on landing that next week.
Comment 14•1 year ago
|
||
Pushed by gtatum@mozilla.com: https://hg.mozilla.org/integration/autoland/rev/9b72a1daed44 Downgrade translations log levels from error to log; r=nordzilla
Comment 15•1 year ago
|
||
bugherder |
Description
•