Hi Frank, Hi Remi,

great thank to both of you and really to Frank for do the work. Good Work!

I can test it on real setup. We will compile and test it. …

Really cool community work. Thanks

Sven
 

Am 01.11.17, 18:10 schrieb "Remi Bergsma" <rberg...@schubergphilis.com>:

    Hi Sven, Frank,
    
    We hit this issue too last year in our CloudStack fork, and this is how we 
resolved it:
    
https://github.com/MissionCriticalCloudOldRepos/cosmic-plugin-hypervisor-kvm/pull/14
    
    Shouldn’t be too hard to port to current CloudStack, but I have no time to 
do it and test it. We have been running like this ever since (18 months) 
without issues.
    
    Regards,
    Remi
    
    
    
    On 01/11/2017, 11:06, "Vogel, Sven" <sven.vo...@kupper-computer.com> wrote:
    
        Hi Frank,
        
        i filed a bug report. I hope anybody can fix that fastly.
        
        https://issues.apache.org/jira/browse/CLOUDSTACK-10127
        
        Greetings
        
        Sven
        
        -----Ursprüngliche Nachricht-----
        Von: Vogel, Sven [mailto:sven.vo...@kupper-computer.com] 
        Gesendet: Mittwoch, 1. November 2017 00:14
        An: dev@cloudstack.apache.org
        Betreff: Re: 4.9 / 4.10 KVM + openvswitch + vpc + static nat / 
secondary ip on eth2?
        
        Hi Frank,
        
        Great analysis, but what can we do now? Is there a workaround or who 
can fix this?
        
        
        Greetings
        
        Sven
        
        
        
        Am 31.10.2017 um 22:41 schrieb Frank Maximus 
<frank.maxi...@nuagenetworks.net<mailto:frank.maxi...@nuagenetworks.net>>:
        
        That seems to be a bug in the lookup of the device number, in case of 
openvswitch.
        The config clearly sets device to eth2, while it should be eth1.
        
        More specifically:
        in LibvirtComputingResource.prepareNetworkElementCommand()
        The broadcastUriToNicNum map is filled depending on the VR nics.
        In openvswitch the guest bridge is used as is, so it overwrites the 
mapping of public.
        This was not an issue until 4.6 as then VR was using the macaddress to 
do lookup, while now it is using the device number.
        
        Kind Regards,
        Frank
        
        
        On Tue, Oct 31, 2017 at 8:35 PM Frank Maximus < 
frank.maxi...@nuagenetworks.net<mailto:frank.maxi...@nuagenetworks.net>> wrote:
        
        I think that the bug you mentioned might not be related.
        Could you please send the content of the file /etc/cloudstack/ips.json 
on the VR.
        
        That might provide useful information.
        
        Kind regards,
        Frank
        
        On Tue, 31 Oct 2017 15:23 Vogel, Sven, 
<sven.vo...@kupper-computer.com<mailto:sven.vo...@kupper-computer.com>>
        wrote:
        
        Hi Devs,
        
        i have the following problem.
        
        When I look this jira ticket I see no solution.
        https://issues.apache.org/jira/browse/CLOUDSTACK-6801  but I think the 
problem is not solved correctly.
        
           1. KVM
           2. Bridges
           bond with two interfaces and trunk (0,129,180,100-1500) to cloudbr0
           Cloudbr0 (0 - guest network)
           Fakebridge pub129 (public network)
           Fakebridge sto180 (secondary storage network)
           Fakebridge  mgmt0 (management)
        
           If I have a vpc all things work until I add a secondary ip and add a 
static nat.
        
           The following will happen, first address will be on the the correct 
interface 146.0.122.134/26 but static nat will be on the false network.
        Its on the eth2.
        
           root@r-29-VM:~# ip a
           1: lo: <LOOPBACK,UP,LOWER_UP> mtu 16436 qdisc noqueue state UNKNOWN
               link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
               inet 127.0.0.1/8 scope host lo
           2: eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast 
state UP qlen 1000
               link/ether 0e:00:a9:fe:03:81 brd ff:ff:ff:ff:ff:ff
               inet 169.254.3.129/16 brd 169.254.255.255 scope global eth0
           3: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast 
state UP qlen 1000
               link/ether 1e:00:2c:00:00:68 brd ff:ff:ff:ff:ff:ff
               inet 146.0.122.134/26 brd 146.0.122.191 scope global eth1
           4: eth2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast 
state UP qlen 1000
               link/ether 02:00:57:07:00:0c brd ff:ff:ff:ff:ff:ff
               inet 192.168.1.254/24 brd 192.168.1.255 scope global eth2
               inet 146.0.122.135/26 brd 146.0.122.191 scope global eth2
        
        Normally I think the secondary ip should be on signed to eth1 not eth2!
        It sets my ip on the guest network vlan range on my cloudbr0 but it 
should be pub129. vnet6 has 1353 guest tag and not the public tag.
        
        [root@kvm01 ~]# ovs-vsctl list-br
        cloud0
        cloudbr0
        mgmt0
        pub129
        sto180
        
        [root@kvm01 ~]# virsh domiflist r-29-VM
        Interface  Type       Source     Model       MAC
        -------------------------------------------------------
        vnet4      bridge     cloud0     virtio      0e:00:a9:fe:03:81
        vnet5      bridge     pub129     virtio      1e:00:2c:00:00:68
        vnet6      bridge     cloudbr0   virtio      02:00:57:07:00:0c
        
           Bridge "cloud0"
               Port "vnet4"
                   Interface "vnet4"
        
               Port "vnet5"
                   tag: 129
                   Interface "vnet5"
               Port "vnet6"
                   tag: 1353
                   Interface "vnet6"
        
        Whats wrong or what can I do to fix this? I hope anybody has an idea.
        
        
        greetings
        Sven Vogel
        Head of Cloud Solutions / Consultants
        Kupper Computer GmbH
        Prager Str. 15
        04103 Leipzig
        
        
        
        
    
    

Reply via email to