> On Feb. 13, 2015, 2:14 a.m., Cong Wang wrote:
> > src/slave/containerizer/isolators/network/port_mapping.cpp, line 1622
> > <https://reviews.apache.org/r/30958/diff/1/?file=862684#file862684line1622>
> >
> >     Unless I am on a wrong tree, why do you create this ingress qdisc again 
> > since we already do it in create()? Even if we really need it, you don't 
> > mention in subject or changelog.
> 
> Chi Zhang wrote:
>     This code is the same as the one in create. If an ingress qdisc already 
> exists, Try will return a some, which is false. Since we only check if Try is 
> an error, we should be fine. This is basically doing 'add them back if none 
> exist now'.
> 
> Cong Wang wrote:
>     I know. So in which case it is "none exist now"? And why do you add it in 
> a patch with this $subject? IOW, why you add some code but hide it in 
> changelog?

The slave won't exit after the last executor but wait for the next container to 
come in, at which point new ingress qdiscs need to be added back. I have 
updated the commit message.


- Chi


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/30958/#review72312
-----------------------------------------------------------


On Feb. 13, 2015, 6:24 p.m., Chi Zhang wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/30958/
> -----------------------------------------------------------
> 
> (Updated Feb. 13, 2015, 6:24 p.m.)
> 
> 
> Review request for mesos, Jie Yu and Cong Wang.
> 
> 
> Repository: mesos
> 
> 
> Description
> -------
> 
> PortMappingIsolator: Better ingress qdisc management on eth0 and lo.
>     
> Remove ingress qdiscs on eth0 and lo after the last container exits. They will
> be added back when the next container comes in. This makes sure after a clean
> shutdown, the sytem will be clean.
> 
> 
> Diffs
> -----
> 
>   src/slave/containerizer/isolators/network/port_mapping.cpp 
> b860b17e59486cd6097183a4d3139fbd5c897b44 
> 
> Diff: https://reviews.apache.org/r/30958/diff/
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> Chi Zhang
> 
>

Reply via email to