[Gluster-devel] Slow regression tests

Jeff Darcy jdarcy at redhat.com
Wed Apr 6 15:40:42 UTC 2016


Lately our times have been all over the map.  I've seen a few runs under
three hours, but also several up around five and a half.  For the longer
runs the slowest tests always seem to be snapshot-related.  This is not
the case for the faster runs, so the problem is not that the snapshot
tests are too slow.  It's that they're *unpredictably* slow some of the
time.  Looking at recent commits doesn't suggest a reason to me.  Does
anyone have any idea why snapshot tests in particular should exhibit
such high run-time variability, or how to fix it?


More information about the Gluster-devel mailing list