-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/15014/
-----------------------------------------------------------
(Updated Oct. 29, 2013, 4:48 p.m.)
Review request for cloudstack, Toshiaki Hatano and Yoshikazu Nojima.
Changes
-------
Updated to validate VNI range isn't used on a different physical network in
zone. Tested with VNI 3000-3500 on guest eth0 and VNI 3500-4000 on guest eth2.
This failed with "VNI 3500 already exists on another network in zone, please
specify a unique range". Changing physical net eth2's range to 3501-4000 was
success. Editing ranges similarly throws the error.
Bugs: CLOUDSTACK-4967
https://issues.apache.org/jira/browse/CLOUDSTACK-4967
Repository: cloudstack-git
Description
-------
1) vxlan will use bridge scheme 'brvx-<vni>'. Multiple physical networks can
host guest traffic type with vxlan isolation, so long as they don't use the
same VNI range.
2) Guest traffic labels can be physical interface if bridge by given name is
not found. Normally we take traffic label name, find the matching bridge, then
resolve that to a physical interface. Then we create guest bridges on that
interface. Now we can just specify the interface.
This patch may need some refinement, but I want to get it to the vxlan devs for
review.
Diffs (updated)
-----
plugins/hypervisors/kvm/src/com/cloud/hypervisor/kvm/resource/BridgeVifDriver.java
f945b74
plugins/hypervisors/kvm/src/com/cloud/hypervisor/kvm/resource/LibvirtComputingResource.java
286d0f7
scripts/vm/network/vnet/modifyvxlan.sh a3ec71f
server/src/com/cloud/network/NetworkServiceImpl.java cf419f3
Diff: https://reviews.apache.org/r/15014/diff/
Testing
-------
I created two physical networks hosting guest traffic, one as a vlan isolation
type, the other a vxlan isolation type. The vlan isolation type has traffic
label 'cloudbr0', on physical eth0, and the vxlan has traffic label 'eth2' on
physical network eth2.
This patch would still allow multiple guest networks to use vxlan isolation,
just not the same VNI numbers. It doesn't enforce this though.
Here they are running side by side:
root@devcloud-kvm-u:~# brctl show
bridge name bridge id STP enabled interfaces
breth0-3905 8000.000c29d82947 no eth0.3905
vnet8
brvx-1213043 8000.a6e6026f7fbb no vnet9
vxlan1213043
brvx-1589169 8000.663d4e875e78 no vnet10
vxlan1589169
cloud0 8000.fe00a9fe0069 no vnet0
vnet2
vnet3
cloudbr0 8000.000c29d82947 no eth0
vnet4
vnet6
cloudbr1 8000.000c29d82951 no eth1
vnet1
vnet5
vnet7
root@devcloud-kvm-u:~/cloudstack# ip -d link show | grep vxlan
56: vxlan1213043: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue
master brvx-1213043 state UNKNOWN mode DEFAULT group default
vxlan id 1213043 group 239.18.130.115 dev eth2 port 32768 61000 ttl 10
ageing 300
59: vxlan1589169: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue
master brvx-1589169 state UNKNOWN mode DEFAULT group default
vxlan id 1589169 group 239.24.63.177 dev eth2 port 32768 61000 ttl 10
ageing 300
62: vxlan1638510: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue
master brvx-1638510 state UNKNOWN mode DEFAULT group default
vxlan id 1638510 group 239.25.0.110 dev eth2 port 32768 61000 ttl 10 ageing
300
root@devcloud-kvm-u:~/cloudstack# ip route show
default via 192.168.100.1 dev cloudbr1 metric 100
169.254.0.0/16 dev cloud0 proto kernel scope link src 169.254.0.1
172.17.10.0/24 dev cloudbr0 proto kernel scope link src 172.17.10.10
192.168.100.0/24 dev cloudbr1 proto kernel scope link src 192.168.100.10
239.18.130.115 dev eth2 scope link
239.24.63.177 dev eth2 scope link
239.25.0.110 dev eth2 scope link
Thanks,
Marcus Sorensen
