[ovs-dev] Troubles with bonding commands
Sara Escribano
sara.escribano.sastre at gmail.com
Mon Nov 19 08:44:35 UTC 2018
Hi everyone, I'm having several troubles with *bonding* functionalities.
I'm beginning in this world and I made a topology like this one on my
virtual machine:
+----+
|ns1 |
+----+
|vpeerns1
|
|
|vns1
+---+ +---+
+----+vpeerns4 vns4 | |vpeerns3 vns3| |vns2
vpeerns2+----+
|ns4 |------------------|br1|----------------------|br0|-------------------|ns2
|
+----+ | | | |
+----+
+---+ +---+
|vpeerns5 - |vns6
| |
| |
| |
| |
| |
+----+
|vns5 |
| | namespaces [nsX]
+---+ |
+----+
+----+vpeerns7 vns7 | | |
+---+
|ns7 |------------------|br2|------------------------|
| |
+----+ | |vpeerns6
| | bridges [brX]
+---+
| |
+---+
(I hope the topology can be understood)
When everything is connected directly to br0, br1 or br2, all devices
are working perfectly, connectivity exists between them and I can make
*ping* between all namespaces.
The problem comes when I want to use *bonding* for vpeerns3 and
vpeerns5 interfaces (forgive the names).
I used these commands to make it (everything worked before this):
*ovs-vsctl add-bond br1 bond1 vpeerns3 vpeerns5 *
*ovs-vsctl set port bond1 bond_mode=balance-tcp*
And now, nothing crosses to ns4 and from ns4. The rest of the pings
are working perfectly (because I set before ARP cache tables) but that
one is not working. The funny thing is when I capture traffic in
vpeerns3/vpeerns5 (I send a ping from ns1 to ns4, for example), there
are echo request messages there, but nothing if I capture on br1, or
ns4, it's like air is between vpeerns3/vpeerns5 and br1.
What could be wrong here??
Thanks in advance
More information about the dev
mailing list