Bugzilla

So the intermittent failure tracked in bug 1572900 has had a low background occurrence rate on a variety of platforms since long before these patches landed, but, looking at the [failures from the day these patches landed and were backed out](https://treeherder.mozilla.org/intermittent-failures.html#/bugdetails?bug=1572900&startday=2020-01-30&endday=2020-01-30&tree=all), they seem to have caused a spike on "windows7-32-mingwclang" -- which is a Tier 2 platform -- in particular.

I did [before](https://treeherder.mozilla.org/#/jobs?repo=try&revision=d78edb284eafd30cf74331a6a0f57e39fda60744) and [after](https://treeherder.mozilla.org/#/jobs?repo=try&revision=954406d38d6b7d07bf2f92b9e6d02bb14af2e58e) try pushes and indeed, reftests on windows7-32-mingwclang have a 65% failure rate with the patches applied (compared to 0%, out of 20 runs for each task), without the patches.

Kats, do you think we should block the re-landing of these patches on figuring out what's going on with those mingw failures? On the one hand, they're Tier 2 failures and these patches fix some user-noticeable regressions. On the other hand, the high failure rate with these patches may be cause for concern.

So the intermittent failure tracked in bug 1572900 has had a low background occurrence rate on a variety of platforms since long before these patches landed, but, looking at the [failures from the day these patches landed and were backed out](https://treeherder.mozilla.org/intermittent-failures.html#/bugdetails?bug=1572900&startday=2020-01-30&endday=2020-01-30&tree=all), they seem to have caused a spike on "windows7-32-mingwclang" -- which is a Tier 2 platform -- in particular.

I did [before](https://treeherder.mozilla.org/#/jobs?repo=try&revision=d78edb284eafd30cf74331a6a0f57e39fda60744) and [after](https://treeherder.mozilla.org/#/jobs?repo=try&revision=954406d38d6b7d07bf2f92b9e6d02bb14af2e58e) try pushes and indeed, reftests on windows7-32-mingwclang have a 65% failure rate with the patches applied (compared to 0%, out of 20 runs for each task), without the patches.

Kats, do you think we should block the re-landing of these patches on figuring out what's going on with those mingw failures? On the one hand, they're Tier 2 failures and these patches fix some user-noticeable regressions. On the other hand, the high failure rate with these patches may indicate a potential cause for concern even on a Tier 2 platform.