Red Hat Bugzilla – Bug 818089
VMs fail to start in case ovirtmgmt bridge doesn't have a link
Last modified: 2013-12-01 04:10:22 EST
RPMs built from git hash ae5318111e935b2f9e3d440283f639d596b0c3e4 installed on Fedora 16.
I disconnected the ethernet cable from the em1 interface which is the only nic under ovirtmgmt bridge, VMs failed to start.
I connected the same nic to a network where there can't be any communication from that nic to other nodes (wrong ip address/netmask) and VMs started successfully.
Steps to reproduce:
1. Create a VM
2. Remove the cable from the ovirtmgmt bridge's interface (maybe ifdown to the bridge might be enough)
3. Start the VM
VDSM logs will be attached later on (in case they won't be override until I get them).
the main issue here is that the host will move in to non-op state once the link is down.
when working with all in one setup local host + local storage we do not have the obligation to have an active link in the host.
(In reply to comment #0)
> VDSM logs will be attached later on (in case they won't be override until I
> get them).
They haven't been...
(In reply to comment #1)
> the main issue here is that the host will move in to non-op state once the
> link is down.
> when working with all in one setup local host + local storage we do not have
> the obligation to have an active link in the host.
If this is the case, it is not a Vdsm bug, but an Engine one.
As a workaround, in current upstream Engine, you can define the network as non-required, and set the news OnlyRequiredNetworksMandatoryForVdsSelection=True
Is this still an issue? I see the reporter did not replied since May.
Any idea how we can get this issue forward or close it?
myakove, could you please help with the needinfo?
It seems that this bug requires some attention at least in AIO setup
I installed AIO setup via
yum install ovirt-engine-setup-plugin-allinone
version: oVirt Engine Version: 3.3.0-0.2.master.20130401204125.gita95f899.fc18
I did ifdown ovirtmgmt && ifdown em1 (underlying interface for ovirtmgmt)
than I connected via another IP to engine GUI (via em2)
shutting down ovirtmgmt resulted in Non Responsive status of local_host (see screenshot 1) which put data-center in non-responsive status so no VM can be started
it seems that there was a problem with connecting to API
ERROR: _get_hypervisors_from_api: [ERROR]::oVirt API connection failure, [Errno 110] Connection timed out
Created attachment 730789 [details]
Created attachment 730793 [details]
Created attachment 730794 [details]
Created attachment 730798 [details]
Created attachment 730806 [details]
Closing old bugs. If this issue is still relevant/important in current version, please re-open the bug.