I have tried the following with 2.2.13 and the new version 3 beta.

Ubuntu host with 4 physical nics, all in one bond0

br50 is management bridged to bond0.50
br60 is storage bridged to bond0.60
br200 is public bridged to bond0.200

Whenever the SSVMs or a VM get deployed to the host somewhere the system tacks on an additional vlan (which is not correct) to the bonded device... i.e.

cloud1:~# brctl show
bridge name bridge id STP enabled interfaces
br50 8000.00e081c5f380 no bond0.50
tap.br50
vnet1
vnet4
br60 8000.00e081c5f380 no bond0.60
brbond0 8000.00e081c5f380 no bond0.200
tap0
tap1
cloud0 8000.fe00a9fe0057 no vnet0
vnet3
vnet7
cloudVirBr200 8000.00e081c5f380 no bond0.200.200
vnet2
vnet5
cloudVirBr500 8000.00e081c5f380 no bond0.50.500
vnet6
virbr0 8000.000000000000 yes


agent.properties:
guest.network.device=br50
private.network.device=br50
public.network.device=brbond0


I can manually:
brctl delif cloudVirBr200 bond0.200.200
brctl addif cloudVirBr200 bond0.200

And the SSVM communicate fine.

If I do public.network.devcie=bond0 it doesn't work with error:
2012-02-17 23:00:56,216{GMT} ERROR [cloud.agent.AgentShell] (main:) Unable to start agent: Failed to get public nic name
2012-02-17 23:00:56,216 ERROR [cloud.agent.AgentShell] (main:null) Unable to start agent: Failed to get public nic name
Unable to start agent: Failed to get public nic name


I would think that specifying bond0 should equate to eth0/1/2/3 and the vlan appended to that device ?



Any pointers?

This looks like a bug to me. I'll be testing bonding and beta3 on CentOS this week, so I'll let you know if it works for me.

In the meantime you can search if this is a know issue (I didn't find it yet) and if not file a bugreport at bugs.cloud.com/

Hi,

Are you using Advanced VLAN mode? If so, that is ho it's supposed to work, but then, different type of networks (public, private, etc.) they have to be on different vlans.

Your system vm's are attaching both to management network and public network (that's how it works in advanced vlan mode), but it looks like they're attaching to different interfaces (bridges) in the same network.

Can you provide your network config in CS?

thx.

Yes. Advanced VLAN mode.

There are different bridges on the OS for the different bonded VLANS, i.e. bond0.50 is bridged to br50.

Vlans in the networking configuration /etc/network/interfaces can be specified with either
VLAN200 or bond0.200 they both do the same tagging on trunk. (which is same as eth0.200 being specified)

I would anticipate/and it would be way easier that in 'advanced mode' using bonded physical interfaces the agents/cloud-stack-configuration should be configured as:

guest.network.device=bond0
private.network.device=bond0
public.network.device=bond0



And the VLAN then is appended to end of that interface (which it tries to do wrongly and only when another bridge device exists) it seems it is doing a lookup somewhere for default route, and when you specify it as above public.network.devcie=bond0 it shouldn't error with:
2012-02-17 23:00:56,216{GMT} ERROR [cloud.agent.AgentShell] (main:) Unable to start agent: Failed to get public nic name
2012-02-17 23:00:56,216 ERROR [cloud.agent.AgentShell] (main:null) Unable to start agent: Failed to get public nic name
Unable to start agent: Failed to get public nic name


This seems that bond0 is not being treated the same as eth0/eth1/eth2 etc... even though it essentially is the same (just not assigned networking IP or other information as that is left for the bridge device to vlan mapping)

Hi!



Hm. As far as I know creating a subinterface (like eth0.N) doesn't make tag vlan. I've used it before to assign more than one IP address on a network interface. I thought vlan stuff is done with vconfig...

thx

I'm not 100% on other linux flavors but ubuntu it is easy as that.

cloud1:~# cat /proc/net/bonding/bond0
Ethernet Channel Bonding Driver: v3.7.1 (April 27, 2011)

Bonding Mode: load balancing (round-robin)
MII Status: up
MII Polling Interval (ms): 100
Up Delay (ms): 200
Down Delay (ms): 200

Slave Interface: eth2
MII Status: up
Speed: 1000 Mbps
Duplex: full
Link Failure Count: 1
Permanent HW addr: 00:e0:81:c5:f3:7e
Slave queue ID: 0

Slave Interface: eth1
MII Status: up
Speed: 1000 Mbps
Duplex: full
Link Failure Count: 1
Permanent HW addr: 00:e0:81:c5:f3:80
Slave queue ID: 0

Slave Interface: eth0
MII Status: up
Speed: 1000 Mbps
Duplex: full
Link Failure Count: 1
Permanent HW addr: 00:e0:81:c5:f3:7f
Slave queue ID: 0

Slave Interface: eth3
MII Status: up
Speed: 1000 Mbps
Duplex: full
Link Failure Count: 1
Permanent HW addr: 00:e0:81:c5:f3:81
Slave queue ID: 0

cloud1:~# cat /proc/net/vlan/bond0.50
bond0.50 VID: 50 REORDER_HDR: 1 dev->priv_flags: 4001
total frames received 269310
total bytes received 17091394
Broadcast/Multicast Rcvd 4952

total frames transmitted 140434
total bytes transmitted 11863358
Device: bond0
INGRESS priority mappings: 0:0 1:0 2:0 3:0 4:0 5:0 6:0 7:0
EGRESS priority mappings: