Failure to start a browser process when running v8.browsing-future benchmarks. |
||||
Issue descriptionThere are several failures because the browser process could not be started. An example log here: https://logs.chromium.org/v/?s=chrome%2Fbb%2Fchromium.perf%2FAndroid_One_Perf%2F938%2F%2B%2Frecipes%2Fsteps%2Fv8.browsing_mobile-future.reference_on_Android%2F0%2Fstdout https://logs.chromium.org/v/?s=chrome%2Fbb%2Fchromium.perf%2FAndroid_One_Perf%2F939%2F%2B%2Frecipes%2Fsteps%2Fv8.browsing_mobile-future.reference_on_Android%2F0%2Fstdout A slightly different trace, but I think the same root cause here: https://logs.chromium.org/v/?s=chrome%2Fbb%2Fchromium.perf%2FAndroid_One_Perf%2F925%2F%2B%2Frecipes%2Fsteps%2Fv8.browsing_mobile-future.reference_on_Android%2F0%2Fstdout
,
Jan 5 2018
,
Jan 5 2018
hablich@, ned@ any ideas who could be the right person to look at this? This seems to be happening quite frequently on reference builds but also sometimes on the non-reference builds.
,
Jan 5 2018
Forgot to mention that these failures are on Android one perf bot.
,
Jan 5 2018
I have confirmed that the first error (failing Start at "android_browser_backend.py") appears to be quite common on reference builds, however I couldn't find any non-reference build with that error. The second error (a crash with error "at java.util.concurrent.FutureTask.finishCompletion") is not as common; but again couldn't find it on any of the most recent 50 non-reference builds. As there isn't much we can do to fix reference builds, I would suggest to close?
,
Jan 5 2018
IIRC, we ignore return code of reference build benchmark for that reason. Unless this happens really frequently to the point that we should update reference build, I would suggest closing this bug as well.
,
Jan 5 2018
I wasn't correct about seeing this failure on non-reference builds. I couldn't find a non-reference build. So closing this bug as suggested. |
||||
►
Sign in to add a comment |
||||
Comment 1 by mythria@chromium.org
, Jan 5 2018