> On July 2, 2014, 5:43 p.m., Ian Downes wrote:
> > Did you consider:
> > hashmap<ContainerID, Option<Info*> > infos
> > This is explicit that we support containers that optionally have network 
> > isolation. Then we can disambiguate so e.g., calling watch() or usage() for 
> > a container that we know about but doesn't have network isolation can be 
> > handled more cleanly and without misleading LOG(WARNING) messages.

we decided to add a separate hashset<ContainerID> unmanaged; for this purpose 
because:
- even more explicit
- infos are used everywhere. a lot to change.


> On July 2, 2014, 5:43 p.m., Ian Downes wrote:
> > src/slave/containerizer/isolators/network/port_mapping.cpp, line 1723
> > <https://reviews.apache.org/r/23214/diff/1/?file=622197#file622197line1723>
> >
> >     This will get called constantly, even for containers that don't use the 
> > network isolator, and thus we'll be logging a lot!
> >     
> >     Suggest removing the log message.

changed to VLOG(1)


> On July 2, 2014, 5:43 p.m., Ian Downes wrote:
> > src/tests/port_mapping_tests.cpp, line 1216
> > <https://reviews.apache.org/r/23214/diff/1/?file=622199#file622199line1216>
> >
> >     Why not just call CreateSlaveFlags again later when you're starting the 
> > second containerizer?
> >     slave::Flags flags = CreateSlaveFlags()
> >     // clear the network isolation parts
> >     // start tasks, etc.
> >     flags = CreateSlaveFlags();

CreatesSlaveFlags creates a new random flags.work_dir every time, so recovery 
wont' kick in since meta dir is somewhere else.


- Chi


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/23214/#review47154
-----------------------------------------------------------


On July 2, 2014, 10:55 p.m., Chi Zhang wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/23214/
> -----------------------------------------------------------
> 
> (Updated July 2, 2014, 10:55 p.m.)
> 
> 
> Review request for mesos, Ian Downes, Jie Yu, and Vinod Kone.
> 
> 
> Bugs: https://issues.apache.org/jira/browse/MESOS-1557
>     
> https://issues.apache.org/jira/browse/https://issues.apache.org/jira/browse/MESOS-1557
> 
> 
> Repository: mesos-git
> 
> 
> Description
> -------
> 
> This eases deployment where a slave could be upgraded to use Network Isolator
> without removing all the existing tasks.
>     
> - Added a new test.
> - Moved all portmapping tests to a new file.
> 
> 
> Diffs
> -----
> 
>   src/Makefile.am 6438939 
>   src/slave/containerizer/isolators/network/port_mapping.hpp ac3bee3 
>   src/slave/containerizer/isolators/network/port_mapping.cpp a326653 
>   src/tests/isolator_tests.cpp 4650f97 
>   src/tests/mesos.cpp 0d7f335 
>   src/tests/port_mapping_tests.cpp PRE-CREATION 
> 
> Diff: https://reviews.apache.org/r/23214/diff/
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> Chi Zhang
> 
>

Reply via email to