cloudstack-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Frank Maximus <frank.maxi...@nuagenetworks.net>
Subject Re: 4.9 / 4.10 KVM + openvswitch + vpc + static nat / secondary ip on eth2?
Date Tue, 31 Oct 2017 21:40:56 GMT
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.maximus@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.Vogel@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
>>
>>
>>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message