[ovs-dev] [PATCH v2 0/5] create tunnel devices using rtnetlink interface

[Thadeu Lima de Souza Cascardo]

On Thu, Jun 09, 2016 at 09:49:15PM -0700, Jesse Gross wrote:
> On Wed, Jun 8, 2016 at 11:21 AM, Thadeu Lima de Souza Cascardo
> <cascardo at redhat.com> wrote:
> > This series adds support for the creation of tunnels using the rtnetlink
> > interface. This will open the possibility for new features and flags on those
> > vports without the need to change vport compatibility code.
> >
> > Support for STT and LISP have not been added because these are not upstream yet,
> > so we don't know how the interface will be like upstream. And there are no
> > features in the current drivers right now we could make use of.
> >
> > We are able to set the MTU to UINT16_MAX since it is not restricted by the
> > driver during newlink.
> 
> The fact that the MTU can be set at link creation time is very nice -
> I was surprised when I looked at the source to verify what was going
> on.
> 
> I did a little bit of testing with this series. For the basic case of
> creating ports that didn't already exist, it worked well. However, it
> had some problems with existing ports. There were two types that I ran
> across:
> 
> * When restarting OVS, it doesn't seem to be able to find and reuse
> the existing tunnel ports. They end up getting removed from the
> datapath and traffic doesn't pass.
> * A little stranger one that I haven't been able to reproduce yet but
> I suspect that relates to old state from a previous version of OVS
> without this patch - I had a GRE port that already existed and when it
> tried to start up it failed with thousands of error messages.
> 
> In the kernel log:
> [4263763.845236] net_ratelimit: 5861 callbacks suppressed
> [4263763.845239] A link change request failed with some changes
> committed already. Interface gre_sys may have been left with an
> inconsistent configuration, please check.
> 
> And in ovs-vswitchd.log:
> 2016-06-10T04:03:19.768Z|01453|dpif|WARN|Dropped 1114 log messages in
> last 1 seconds (most recently, 0 seconds ago) due to excessive rate
> 2016-06-10T04:03:19.768Z|01454|dpif|WARN|system at ovs-system: failed to
> add tun0 as port: File exists

I guess I need the previous line that was dropped, or something referring to
gre_sys instead of tun0.

> 
> Can you take a look at these and also maybe do some more testing in
> these types of edge scenarios?

I will. On the previous version, I remember testing an upgrade from OVS without
the patch to OVS with the patch. I will look into those issues and test both a
restart and an upgrade scenario on some different kernels.

Thanks.
Cascardo.