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


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.


src/slave/containerizer/isolators/network/port_mapping.cpp
<https://reviews.apache.org/r/23214/#comment82740>

    s/PID/pid/
    
    change language to say this is *possibly* because it's a container that 
wasn't managed by the network isolator



src/slave/containerizer/isolators/network/port_mapping.cpp
<https://reviews.apache.org/r/23214/#comment82741>

    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.



src/tests/isolator_tests.cpp
<https://reviews.apache.org/r/23214/#comment82841>

    This doesn't appear to be used in this file?



src/tests/port_mapping_tests.cpp
<https://reviews.apache.org/r/23214/#comment82845>

    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();



src/tests/port_mapping_tests.cpp
<https://reviews.apache.org/r/23214/#comment82844>

    s/Reset/Clear/



src/tests/port_mapping_tests.cpp
<https://reviews.apache.org/r/23214/#comment82842>

    s/islator/isolator/



src/tests/port_mapping_tests.cpp
<https://reviews.apache.org/r/23214/#comment82846>

    just name the variable recover?



src/tests/port_mapping_tests.cpp
<https://reviews.apache.org/r/23214/#comment82847>

    Add some commenting about why you need three slaves:
    1) Start task without network isolator
    2) Recover task with slave and run new task with network isolator
    3) Recover tasks where one has network isolation and one does not.


- Ian Downes


On July 1, 2014, 2:22 p.m., Chi Zhang wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/23214/
> -----------------------------------------------------------
> 
> (Updated July 1, 2014, 2:22 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 e3ff6d7 
>   src/slave/containerizer/isolators/network/port_mapping.cpp a326653 
>   src/tests/isolator_tests.cpp 4650f97 
>   src/tests/port_mapping_tests.cpp PRE-CREATION 
> 
> Diff: https://reviews.apache.org/r/23214/diff/
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> Chi Zhang
> 
>

Reply via email to