[lxc-devel] [PATCH] clarify explanation of veth

[Stéphane Graber]

On Fri, Jan 16, 2015 at 04:50:50PM +0100, Guido Jäkel wrote:
> On 16.01.2015 16:31, Stéphane Graber wrote:
> > On Fri, Jan 16, 2015 at 04:13:39PM +0100, Guido Jäkel wrote:
> >> On 15.01.2015 23:08, Stéphane Graber ACKd:
> >>> +              By default, <command>lxc</command> chooses a name for the
> >>
> >> Is this the truth? I expect the veth driver (or the system) itself create the default names.
> >>
> >>
> >> BTW: I'm using the container name as the outerside name for the veth. For me, is very convenient to have this direct correlation, e.g. for automated naming at SNMP-Discovery and -Monitoring.
> >>
> >> Guido
> > 
> > It's the truth. LXC uses a vethXXXX template rather than what you'd get
> > from the kernel which is just vethX where X is an incrementing digit.
> 
> Well, I should have taken a look in the code repository before. Please, may I then (nevertheless) ask for the motivation of the choice for pattern vethXXXX ?

I'm not sure what the initial reason for it is. However to me it makes
it trivial to differentiate lxc-created veth devices vs manually created
ones.

Another potential reason why this was done to begin with is that if they
are sequentially numbered and a container side veth is moved back to the
host namespace, a name clash is likely which would then make the kernel
rename it to renameX and would confuse anything which assumes that veth*
are veth devices and that everything else can be used for host
connectivity (like NetworkManager).


-- 
Stéphane Graber
Ubuntu developer
http://www.ubuntu.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: Digital signature
URL: <http://lists.linuxcontainers.org/pipermail/lxc-devel/attachments/20150116/d9f66907/attachment.sig>