> On June 26, 2014, 6:29 p.m., Vinod Kone wrote: > > src/tests/isolator_tests.cpp, lines 895-906 > > <https://reviews.apache.org/r/22471/diff/5-6/?file=616523#file616523line895> > > > > Could these be setup by the operators on the host irrespective of the > > test? If yes, removing them seems dangerous?
In theory we probably shouldn't do this, as eth0 and lo are indeed shared resources of the host, but we've in effect taken full control of the network resources anyway, such as host ephemeral range, etc. The isolator is fine even if we don't remove the qdiscs here and I am fine with that. The only concern I can think of right now is these qidcs could have fiters with port ranges that overlap with the ones we are going to add. Things might fail later in a more obscure way. - Chi ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/22471/#review46771 ----------------------------------------------------------- On June 26, 2014, 7:05 p.m., Chi Zhang wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/22471/ > ----------------------------------------------------------- > > (Updated June 26, 2014, 7:05 p.m.) > > > Review request for mesos, Ian Downes, Jie Yu, Vinod Kone, and Cong Wang. > > > Repository: mesos-git > > > Description > ------- > > These tests test different protocols (TCP, UDP, ICMP, ARP, DNS) in different > connection scenarios (C2C, H2C). > > > Diffs > ----- > > src/tests/isolator_tests.cpp 6fef4e0 > > Diff: https://reviews.apache.org/r/22471/diff/ > > > Testing > ------- > > make check. > > > Thanks, > > Chi Zhang > >
