New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Safari stable runs produce duplicate, conflicting results for some tests #20607
Comments
Checking with vim, there are two entries for
|
This is the first time I've seen this. @Hexcles have you ever come across it? My initial guess for how this could happen is if the manifest generated in each job isn't identical, but even then hashing should be by test name... |
Oh wait, the duplication was in a single report. Then I don't have a guess. |
The message "TestRunner hit external timeout (this may indicate a hang)" was added in #20321 so adding @jgraham @LukeZielinski who were on that PR in addition to @Hexcles. Was it expected from those changes that it could result in duplicate entries in the report? |
According to stackdriver, duplicate "/webrtc/RTCRtpParameters-rtcp.html", was first seen 8 days ago (Nov 26th). There have been 8 instances in that time. Conflicting 'browser_name' was first seen on the 25th June 2019. There have been 2 instances in the last 30 days. |
@stephenmcgruer do we have an easy way to see when the browser or OS version has changed in the history of runs? |
This looks like a race condition related to #20321 , but I'm not sure why it only happens to that particular test. Note that the duration of the second entry isn't even close to timeout. |
That's an Edge Canary run. Filed #20613. |
Actually, this is happening to other tests as well. Here's a list of tests with conflicting results we've seen so far:
|
Offhand I'm not sure; what is your underlying thought here @foolip? |
Agreed, I think we want to come up with a fix instead. Do we understand yet what the race condition is? ('No' is a reasonable answer here!) |
I just wanted to check if there was a change in browser or OS version around the time this started happening. |
I did not, but after some investigation today I have a clue.
Two things stood out:
cc @jgraham |
There is no Safari stable run for December 3rd, 2019 (https://wpt.fyi/runs?label=master&label=stable&from=2019-12-02&to=2019-12-05).
Looking at azure pipelines, the build for epochs/daily for that day was https://dev.azure.com/web-platform-tests/wpt/_build/results?buildId=37728
wpt.fyi shows that as INVALID in the status page (having grabbed the run id from internal logs): https://wpt.fyi/status
Checking internal server logs, seems there are two errors that were logged on Dec 3rd:
and
The webrtc conflict was seen 3 times on Dec 3rd, the browser_version conflict once.
The results link for this build was https://dev.azure.com/web-platform-tests/b14026b4-9423-4454-858f-bf76cf6d1faa/_apis/build/builds/37728/artifacts?artifactName=safari-results&api-version=5.1&%24format=zip, I'm going to check now to see which of the two happened to this report.
The text was updated successfully, but these errors were encountered: