> On March 22, 2016, 11:37 p.m., Avinash sridharan wrote: > > src/slave/containerizer/mesos/isolators/network/cni/cni.hpp, line 151 > > <https://reviews.apache.org/r/45082/diff/1/?file=1307516#file1307516line151> > > > > I guess if we are going to use `attach` for connecting to networks, > > lets use `detach` over here.
Agree. > On March 22, 2016, 11:37 p.m., Avinash sridharan wrote: > > src/slave/containerizer/mesos/isolators/network/cni/cni.cpp, lines 647-649 > > <https://reviews.apache.org/r/45082/diff/1/?file=1307517#file1307517line647> > > > > This is a bit odd, __disconnect always returns an error ? The plugin > > can return error codes and error logs which we should be propagating > > upstream through failure semantics. The reason that `__disconnect()` always returns a `Failure` is, it will be only called when the exit code of plugin is not 0 (i.e., plugin fails for a reason), in case of plugin success, we will return `Nothing()` in `_disconnect()`. So we only need `__disconnect()` in case of plugin failure to get the output (i.e., detailed failure reason) of plugin and propagate it upstream. > On March 22, 2016, 11:37 p.m., Avinash sridharan wrote: > > src/slave/containerizer/mesos/isolators/network/cni/cni.cpp, line 653 > > <https://reviews.apache.org/r/45082/diff/1/?file=1307517#file1307517line653> > > > > Why are we returning a Future<list<Nothing>> over here? We should be > > returning a Future<Nothing> there is no list since you are deletnig the > > entire container directory in one shot. This seems a bit odd. I tend to agree with you, however, I see CPU share isolator does the similar thing: https://github.com/apache/mesos/blob/0.28.0/src/slave/containerizer/mesos/isolators/cgroups/cpushare.cpp#L536:L547, maybe we should change it as well? > On March 22, 2016, 11:37 p.m., Avinash sridharan wrote: > > src/slave/containerizer/mesos/isolators/network/cni/cni.cpp, lines 644-645 > > <https://reviews.apache.org/r/45082/diff/1/?file=1307517#file1307517line644> > > > > These two CHECKS don't make sense. What if the plugin got deleted or > > there was a bug in the plugin because of which it wasn't able to delete the > > interfaces or release the IP addresses. The Agent should not die because of > > an error in a 3rd party plugin. If the plugin got deleted, then `Subprocess.isError()` will be true and we will return a `Failure`, if there was a bug in the plugin, then `Subprocess.status()` will be non-zero and we will read its output for the detailed error message. For either of these cases, agent will not die. - Qian ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/45082/#review124806 ----------------------------------------------------------- On March 21, 2016, 12:30 a.m., Qian Zhang wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/45082/ > ----------------------------------------------------------- > > (Updated March 21, 2016, 12:30 a.m.) > > > Review request for mesos, Avinash sridharan, Gilbert Song, and Jie Yu. > > > Bugs: MESOS-4759 > https://issues.apache.org/jira/browse/MESOS-4759 > > > Repository: mesos > > > Description > ------- > > Implemented cleanup() method of "network/cni" isolator. > > > Diffs > ----- > > src/slave/containerizer/mesos/isolators/network/cni/cni.hpp PRE-CREATION > src/slave/containerizer/mesos/isolators/network/cni/cni.cpp PRE-CREATION > > Diff: https://reviews.apache.org/r/45082/diff/ > > > Testing > ------- > > make check > > > Thanks, > > Qian Zhang > >
