----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/32660/#review78564 -----------------------------------------------------------
I feel if we use the native libnl api, we could shift a lot of the testing effort to just verifying the library. The test there could be simpler since you don't have to construct the 'mesos cluster' but rather directly verify the link stats after network manipulation (the iperf interface is a very useful addition to generl network testing in the code base!) Then on the testing of port mapping isolator side, we'd only have to care about 'isolator level' stuff such as flag combos, verify that 'usage' code path is through etc. Also I feel like the coding / testing effort here is pretty signifant. Would it be useful to document on the ticket to help achieve some high-level consensus before diving right into it? - Chi Zhang On April 1, 2015, 3:45 p.m., Paul Brett wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/32660/ > ----------------------------------------------------------- > > (Updated April 1, 2015, 3:45 p.m.) > > > Review request for mesos, Chi Zhang, Ian Downes, and Cong Wang. > > > Bugs: mesos-2332 > https://issues.apache.org/jira/browse/mesos-2332 > > > Repository: mesos > > > Description > ------- > > Report network isolator statistics on a per container basis (MESOS-2332) > > > Diffs > ----- > > include/mesos/mesos.proto 3c592d5ab3092ecbeddfaff95e0c1addc3ac58f8 > src/slave/containerizer/isolators/network/port_mapping.hpp > 33837b4662959a003c8f38d1e786c6615287a4ff > src/slave/containerizer/isolators/network/port_mapping.cpp > e691d463515084518c94cdec3fbdf37be4a72977 > src/slave/flags.hpp 3da71afad38ae41adefab979dbed2ae0b10a98ef > > Diff: https://reviews.apache.org/r/32660/diff/ > > > Testing > ------- > > make check > > > Thanks, > > Paul Brett > >
