Frequent patch failures on CQ runs |
|||
Issue descriptionI'm getting frequent patch failure errors on a large patch (affecting 40 files): https://chromium-review.googlesource.com/c/545089/ For example, patchset 9: Failed to apply patch for presubmit: https://build.chromium.org/p/tryserver.chromium.linux/builders/chromium_presubmit/builds/473230 Succeeded to apply patch for linux_chromium_rel_ng: https://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_rel_ng/builds/487402 This seems to be flaky because a second run of the presubmit succeeds: https://build.chromium.org/p/tryserver.chromium.linux/builders/chromium_presubmit/builds/473236
,
Jun 28 2017
Looks like a flake indeed. Normally these errors would be caused by HEAD being different for the builds and thus patch applying differently, but in this case it looks like both builds have fetched chromium/src at 96f653f3e3ff. One difference I've noticed is that the location of the checkout was different between bots: - /b/build/slave/linux/build/src - /b/c/b/linux/src The error message for failed patch application was: error: Your local changes to the following files would be overwritten by merge: third_party/WebKit/Source/web/tests/data/mhtml/simple_test.mht Full log here: https://luci-logdog.appspot.com/v/?s=chromium%2Fbb%2Ftryserver.chromium.linux%2Fchromium_presubmit%2F473230%2F%2B%2Frecipes%2Fsteps%2Fbot_update%2F0%2Fstdout. Not sure why "git reset --hard" and "git clean -dff" didn't wipe these local changes. Besides, patch does not touch this file, so not sure why would any changes in this file would be overridden... perhaps this is related to how rebase command works. Adding to trooper queue to find owners.
,
Jun 28 2017
We had some issues with crlf characters in that file. I was able to remove that file from the change in a later patchset. I wonder if the CRLF line ending change in third_party/WebKit/Source/web/tests/data/mhtml/simple_test.mht caused these failures?
,
Jul 1 2017
Is this still an issue? The last flakes were reported 6 days ago it seems.
,
Jul 1 2017
It's hard to say what caused this bug because there were several things going on: odd line ending, large patch, and possibly infra issues at the time. I don't think this is really actionable anymore. |
|||
►
Sign in to add a comment |
|||
Comment 1 by mmoss@chromium.org
, Jun 26 2017