Thanks Eric. I also see some intermittent test failures in recent builds.
Maybe we can mark those failing tests as @Flaky, and/or add the RetryRule() to them. Thoughts? On Tue, Sep 1, 2020 at 10:48 AM Eric Badger <ebad...@verizonmedia.com.invalid> wrote: > While putting up patches for HADOOP-17169 > <https://issues.apache.org/jira/browse/HADOOP-17169> I noticed that the > unit tests in trunk, specifically in HDFS, are incredibly unstable. Every > time I put up a new patch, 4-8 unit tests failed with failures that were > completely unrelated to the patch. I'm pretty confident in that since the > patch is simply changing variable names. I also ran the unit tests locally > and they would pass (or fail intermittently). > > Is there an effort to stabilize the unit tests? I don't know if these are > bugs or if they're bad tests. But in either case, it's bad for the > stability of the project. > > Eric > -- L
