> On Dec. 15, 2014, 6:51 p.m., Jie Yu wrote:
> > include/mesos/mesos.proto, line 483
> > <https://reviews.apache.org/r/29033/diff/1/?file=791611#file791611line483>
> >
> >     Ditto for removing the comments.
> >     
> >     s/net_tcp_total_tw_connections/net_tcp_time_wait_connections/

changed the variable names, but I kept the comments -- no harm to have some 
documentation where they are first introduced?


> On Dec. 15, 2014, 6:51 p.m., Jie Yu wrote:
> > src/slave/flags.hpp, lines 375-377
> > <https://reviews.apache.org/r/29033/diff/1/?file=791614#file791614line375>
> >
> >     I would suggest using two separate flags:
> >     
> >     ```
> >     network_enable_socket_statistics_details
> >     network_enable_socket_statistics_summary
> >     ```
> >     
> >     The benefit is that you can choose to turn off all of them (which we 
> > will use initially).

good point. fixed.


> On Dec. 15, 2014, 6:51 p.m., Jie Yu wrote:
> > src/slave/containerizer/isolators/network/port_mapping.cpp, lines 2202-2227
> > <https://reviews.apache.org/r/29033/diff/1/?file=791613#file791613line2202>
> >
> >     Why doing this change? I would suggest keep the original code because 
> > it implies a ALL-OR-NOTHING semantics for those rtt statistics.

1. I intended for only `PortMappingStatistics::execute` to control its stats 
collection / output though the use of flags.
2. `::__usage` can then be simplied try to just get all the results, ignoring 
the flags, otherwise, the condition logic based on the flags needs to be 
repeated here as well, e.g, to do all_or_nothing for the RTT values one needs 
checking the 'expensive' flag is true.
3. a new test has been added that keeps the all-or-nothing check.


- Chi


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/29033/#review65101
-----------------------------------------------------------


On Dec. 23, 2014, 7:51 p.m., Chi Zhang wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/29033/
> -----------------------------------------------------------
> 
> (Updated Dec. 23, 2014, 7:51 p.m.)
> 
> 
> Review request for mesos, Dominic Hamon, Ian Downes, Jie Yu, and Cong Wang.
> 
> 
> Repository: mesos-git
> 
> 
> Description
> -------
> 
> While we are still working to make RTT less expensive to get, expose total 
> number of tcp connections in use and in TIME_WAIT. 
> 
> 
> Diffs
> -----
> 
>   include/mesos/mesos.proto 540071d 
>   src/slave/containerizer/isolators/network/port_mapping.hpp f1e2352 
>   src/slave/containerizer/isolators/network/port_mapping.cpp 2d81336 
>   src/slave/flags.hpp 670997d 
>   src/tests/port_mapping_tests.cpp eb82993 
> 
> Diff: https://reviews.apache.org/r/29033/diff/
> 
> 
> Testing
> -------
> 
> expanded a test case to test PortMappingStatistics.
> 
> 
> Thanks,
> 
> Chi Zhang
> 
>

Reply via email to