You probably need to look at the logic of the tests. Some of these expect a
certain number of files or records in each file and that may not always be the
case in the first or last interval.
Ralph
> On Aug 26, 2020, at 12:47 AM, Volkan Yazıcı wrote:
>
> I have been trying to fix certain (rand
I always prefer to refactor sleep calls to use count down latches or other
concurrency mechanisms due to timing variability like this. Do note that
the Clock instance is configurable, so perhaps a test version can be used
for that test to artificially advance time step by step?
On Wed, Aug 26, 202
I have been trying to fix certain (randomly?) failing tests on both master
and release-2.x; RollingDirectSizeTimeNewDirectoryTest,
RollingDirectTimeNewDirectoryTest, etc. to name a few. I suspect the
hardwiring to the system clock causes this strange behaviour. Consider the
following RollingDirect
