Follow-up on the "egress qos" failure (ovn tests 85/86). This is a testbed/environment issue.
The test greps for a fixed `cburst 3.....b` in the `tc` output, but that value depends on the kernel timer resolution of the machine running the test (via the 4th field of /proc/net/psched = NSEC_PER_SEC / hrtimer_resolution): With high-resolution timers (e.g. upstream CI) -> cburst ~= 374933b -> matches "3.....b" -> test PASSES With coarse timers (e.g. virtualized autopkgtest testbeds) -> cburst is 8 digits -> does not match -> test FAILS This is why it fails on the autopkgtest infrastructure (both amd64 and s390x) but was never seen upstream. It is independent of the openvswitch version (the functions that compute the HTB burst/cburst have not been touched since OVS 3.2.0) and not a regression from 3.3.9. -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/2147329 Title: [SRU] openvswitch 3.3.9 point release To manage notifications about this bug go to: https://bugs.launchpad.net/cloud-archive/+bug/2147329/+subscriptions -- ubuntu-bugs mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
