the i386 failure should be transient. think I can see how that would happen, and can make it less likely by just bumping the sleep, since if this test is running on a single core bad prioritization could have the internal timer fail to update in time.
the i386 failure should be transient. think I can see how that would happen, and can make it less likely by just bumping the sleep, since if this test is running on a single core bad prioritization could have the internal timer fail to update in time.