Hello Sebastian,

Cloudstack creates a new bridge because it's standard behaviour if you mention 
a VLAN number when you add the public network details.

I believe you can just skip mentioning the VLAN id when you add the public 
network, in which case it will just use the bridge you created manually 
(cloudbr1) - I have never done this but there's a good chance it will work.

What I would do in your shoes is just add bond0 (ie no VLAN) to cloudbr1 and 
let Cloudstack build the interfaces automatically as required (bond0.83 and the 
respective bridge). This would be closer to recommended practice.


HTH
Lucian

--
Sent from the Delta quadrant using Borg technology!

Nux!
www.nux.ro

----- Original Message -----
> From: "Sebastian Gomez" <tioc...@gmail.com>
> To: users@cloudstack.apache.org
> Sent: Thursday, 28 January, 2016 12:51:08
> Subject: cloudstack-agent 4.5 problem with cloudbr bridges

> Hi all,
> 
> I have been working with cloudstack and vmware for more than 3 years, and
> now I would like to test it with KVM, but I have problems with the bridges
> names.
> 
> Summarizing, when I add the KVM host on cloudstack, the cloudstack agent
> creates a new bridge a part of the cloudbr0 and cloudbr1 and gets an error
> trying to add the interface associated with the public network VLAN.
> 
> Extended:
> 
> I followed many how-tos, each different, and none of them took me to solve
> the problem.
> 
> I want to configure the KVM and I have these networks:
> 
> 172.16.1.0/24 for management traffic, VLAN 554
> 10.4.0.0/24 for public traffic, VLAN 83
> 10.254.0.0/24 for Guest NW, VLANs [555 - 569]
> 
> Storage NFS is defined under management NW (172.16.1.0/24).
> 
> The KVM host have 2 physical interfaces, configured as a bonding -rr (the
> switch is configured to allow bonding). This is the interface bond0.
> 
> Over it I created:
> - bond0.554 is a virtual interface, configured with the management VLAN
> (554)
> - bond0.83 is a virtual interface, configured with the public VLAN (83)
> 
> And over them, 2 bridges as Cloudstack docs says:
> - cloudbr0 is going to host management traffic and
> - cloudbr1 is going to host public and guest traffic.
> 
> Results in:
> [root@gary1 network-scripts]# brctl show
> bridge name     bridge id               STP enabled     interfaces
> br0             8000.180373f5a953       no              bond0.197
> cloudbr0                8000.180373f5a953       no              bond0.554
> cloudbr1                8000.180373f5a953       no              bond0.83
> 
> The routes are:
> [root@gary1 network-scripts]# route -n
> Kernel IP routing table
> Destination     Gateway         Genmask         Flags Metric Ref    Use
> Iface
> 147.83.197.0    0.0.0.0         255.255.255.0   U     0      0        0 br0
> 10.4.82.0       0.0.0.0         255.255.255.0   U     0      0        0
> cloudbr1
> 172.16.1.0      0.0.0.0         255.255.255.0   U     0      0        0
> cloudbr0
> 192.168.122.0   0.0.0.0         255.255.255.0   U     0      0        0
> virbr0
> 0.0.0.0         147.83.197.1    0.0.0.0         UG    0      0        0 br0
> 
> And I can ping other elements into each network.
> 
> Now, with the agent fresh installed (free of configurations), I add the
> host on cloudstack (4.5.2), and the process stop with an error informing
> that the iface "bond0.83" can't be attached to a new bridge that it created
> "brbond0-83".
> 
> I can see on the host that there is a new bridge called "brbond0-83",
> associated with NO interfaces:
> 
> [root@gary1 network-scripts]# brctl show
> bridge name     bridge id               STP enabled     interfaces
> br0             8000.180373f5a953       no              bond0.197
> brbond0-83              8000.000000000000       no
> cloud0          8000.fe00a9fe00b5       no
> cloudbr0                8000.180373f5a953       no              bond0.554
> cloudbr1                8000.180373f5a953       no              bond0.83
> virbr0          8000.525400244916       yes             virbr0-nic
> 
> Now, I manually detach bond0.83 from cloudbr1 and attach it to the new
> created bridge (brbond0-83), and the process continues.
> 
> The problem is that if I restart the server, the agent cannot re-attach the
> iface again, I have to do it by hand (not good). Also, if I configure the
> system not to attach bond0.83 to the bridge, it generates an error about
> that the network is not found...
> 
> On the cloudstack management service, I defined two physical networks:
> 
> Physical network                Traffic types       KVM Traffic label
> cloudbr0-CS-MGMT            Management      cloudbr0
> cloudbr1-CS-PUB-GUEST   Public, Guest     cloudbr1
> 
> 
> 
> 
> My question is:
> 
> Why cloudstack creates a new bridge?
> 
> 
> 
> Thanks in advanced.

Reply via email to