[ovs-discuss] NFS Xen and OVS

[Florian Manschwetus]

Moin,
I have currently trouble with nfs from dom0 to guest.

I use debian testing originally with ovs-1.4.2 (whatever shipped with 
testing), also xen kernel etc.
As nfs server I use a nexenta community on xen (managed by ganeti using 
drbd)

Due to the issues I build ovs-1.9.0 debs and updated my systems, but 
nothing has changed. (found some backport problem in etherdevice.h 
3.2.41 needs to be excluded also).

My Server has two gbit links one directly to its neighbor (drbd link)
for practical reasons I use both links a uplinks for a bridge (br0 main, 
br1 internal net)
and added the local ports I use as internal ports. (veth0, master0 at 
br0 and drbd0 at br1)
Using this from another host all is fine and the speeds are normally ok.
But when I try to copy some larger files for example with cp or rsync 
from dom0 to automounted nfs (here some tarred Linux images I use to 
setup systems (vms and physical)). The Network freezes totally (no ping 
no ssh nothing), but the ssh shell I have had open issuing the command 
or some other having activity stay active for some time. But it is 
impossible to reboot the server using reboot -f, maybe nosync would do.
Also there is no way to recover networking as far as I tested it.

Is it possibly an packetsize issue? as I have rsize and wsize at 32768?

Would be great if someone has a further idea about this issue and about 
how to solve it.

Regards,
Florian