Search Mailing List Archives


Limit search to: Subject & Body Subject Author
Sort by: Reverse Sort
Limit to: All This Week Last Week This Month Last Month
Select Date Range     through    

[mininet-discuss] ovs-vswitchd support?

Bob Lantz rlantz at cs.stanford.edu
Fri Oct 28 05:06:20 PDT 2011


On Oct 28, 2011, at 4:08 AM, Parantapa Bhattacharya wrote:

>>>> 
>>>> Using the multiprocessing module is interesting. I didn't use call() since it fails with large numbers of file descriptors.
>>> 
>>> Did you try close_fds parameter for call? Anyway i didn't hit the
>>> limit so it didn't bother me.
>> 
>> I do that, but the issue is with the parent, not the child.
> 
> Maybe you were opening files needed only by the child in the parent.
> Need to check on that.

It's when you make 4096 hosts and have pipes to them all.

> 
>> 1. Perhaps my wording was unclear - by "pick" I did not mean "load" but rather "select or identify", that is, how might you *identify the library's name* in a cross-platform, non-hard-wired way?
>> 
> 
> Hmm. Truly speaking I never thought of that. However looking at
> unshare(2) and libc(7) man page, it seems that unshare is a Linux only
> system call.
> Also de facto standard library, glibc uses libc.so.6 to point to the
> actual libc dynamic library. I wont bother much until i face a case
> where someone has a problem.

By cross-platform I meant cross-linux, although BSD does support similar features.

> 
>> 2. That doesn't answer the question: why include support for the linux bridge if you don't need to?
>> 
> 
> What I wrote was truly a one weekend thing :P. I wanted to get started
> quick and dirty. Figuring out how to do the stuff was more interesting
> to me during those 2 days than to write a proper mininet alternative.
> Thus bridge stayed. Moreover, linux bridge class is more or less
> negligible and basically served as my interface for the OpenVSwitch
> class. It is not also tied to any other part of the code. So if anyone
> doesn't like it just delete it :P. And by small i really meant without
> any proper mininet like shell and the other ui related stuff.
> 
>> It's easy to bridge to the root namespace (see hwintf.py and sshd.py) -
>> 
> 
> I agree on that. But it depends on having interfaces properly
> configured before you can start your forwarding. I wanted something
> that doesn't depend on that. I wanted a solution where I should be
> able to configure and my apps shouldn't depend on the vswitches and
> veths.

Advantages and disadvantages. If you want to be able to ssh into your VMs, then they need to be configured somehow. I simply noted that it's not as difficult as one might imagine to do so.

> 
>> Parametrized topologies are extremely useful; a library of topologies is also extremely useful; neither one prevents you building your own. One might also argue that having a mapping between graphs which can be computed upon, plotted, etc. is also useful.
>> 
> 
> Firstly it was my opinion. I want things small simple and separate.
> And really, how difficult is it to attach hosts, switches, and links
> to a graph once you have created the graph ? Anyway, I don't think
> there is any point in arguing about opinions. Authors should have the
> choice to write stuff the way they want.

I was explaining the rationale. I'm fine with multiple API layers however.

> 
> -- 
> Parantapa Bhattacharya



More information about the mininet-discuss mailing list