[
https://issues.apache.org/jira/browse/CLOUDSTACK-9811?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15937100#comment-15937100
]
ASF GitHub Bot commented on CLOUDSTACK-9811:
--------------------------------------------
Github user ustcweizhou commented on the issue:
https://github.com/apache/cloudstack/pull/2011
@DaanHoogland @karuturi @borisstoyanov
Here is a short explanation why the issue happened:
Normally getBroadcastUriFromBridge method in KVM plugin will get the URI
from a bridge name. for example, breth1-100 -> vlan://100, brem1-200 ->
vlan://200.
However, for the bridge like brp55p1-100, it will return vlan://1 which is
obviously wrong.
The method getBroadcastUriFromBridge is used in
prepareNetworkElementCommand for command IpAssocCommand in VR. Each time when
a new ip is associated, it will check the existing nics by getting the URIs
from bridge name of nics. If the URIs do not contain the uri of te new ip, then
a new nic will be plugged and new ip will be associated to the new nic.
The issue happened when cloudstack try to set more information of eth2
(public nic) of a VR, as for the error described in part 1 , the URI of eth2
return wrong result (vlan://1) which is different from uri of ip (it is
vlan://7 from StartCommand), so eth3 is added.
> VR will not start, looking to configure eth3 while no such device exists on
> the VR. On KVM-CentOS6.8 physical host
> ------------------------------------------------------------------------------------------------------------------
>
> Key: CLOUDSTACK-9811
> URL: https://issues.apache.org/jira/browse/CLOUDSTACK-9811
> Project: CloudStack
> Issue Type: Bug
> Security Level: Public(Anyone can view this level - this is the
> default.)
> Components: Virtual Router
> Affects Versions: 4.10.0.0
> Reporter: Boris Stoyanov
> Priority: Blocker
> Attachments: agent.log, cloud.log, management.log
>
>
> This issue appears only on 4.10. When you add an instance with a new network
> the VR starts and fails at the configuration point. Looks like it is looking
> to configure eth3 adapter while no such device should be available on the VR.
> The VR does not start and aborts the deployment of the VM.
> Pease note that this issue was reproduced on physical KVM hosts in our lab.
> Hardware Hosts details:
> - 4x Dell C6100
> - Using: American Megatrends MegaRAC Baseboard Management (IPMI v2 compliant)
> OS:
> CentOS 6.8.
> Management:
> VM, running CentOS 6.8
> ACS version: 4.10 RC 1. SHA: 7c1d003b5269b375d87f4f6cfff8a144f0608b67
> In a nested virtualization environment it was working fine with CentOS6.8.
> Attached are the management log and the cloud.log form the VR.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
