----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/39531/#review103496 -----------------------------------------------------------
Should we also mention it explicitly that each NetworkInfo message represents a request to create a new virtual network interface inside the network namespace for the Container? As a separate note, we also need to update the proto files inside include/v1 :-). include/mesos/mesos.proto (line 1382) <https://reviews.apache.org/r/39531/#comment161549> Should we call it IPAddress instead? - Kapil Arya On Oct. 21, 2015, 8:35 p.m., Connor Doyle wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/39531/ > ----------------------------------------------------------- > > (Updated Oct. 21, 2015, 8:35 p.m.) > > > Review request for mesos, Benjamin Hindman and Kapil Arya. > > > Bugs: MESOS-3788 > https://issues.apache.org/jira/browse/MESOS-3788 > > > Repository: mesos > > > Description > ------- > > In Mesos 0.25.0, a new message called NetworkInfo was introduced. This > message allows framework authors to communicate with network isolation > modules via a first-class message type to request IP addresses and network > group isolation policies. > > Unfortunately, the structure is somewhat confusing to both framework authors > and module implementors. > > 1) It's unclear how IP addresses map to virtual interfaces inside the > container. > 2) It's difficult for application developers to understand the final policy > when multiple IP addresses can be assigned with differing isolation policies. > > NOTE: the slave should also be updated in order to set the new IpAddress > message in ContainerStatus in case network isolation modules are not > installed. > > > Diffs > ----- > > include/mesos/mesos.proto 4a16be1 > > Diff: https://reviews.apache.org/r/39531/diff/ > > > Testing > ------- > > make && make check > > > Thanks, > > Connor Doyle > >
