I don't think it's necessary.
Please provide the host's routing table and interfaces list ('ip a' or ifconfing) while it's configured with the bond.

Thanks

On Tue, Aug 16, 2016 at 4:39 PM, Uwe Laverenz <uwe@laverenz.de> wrote:
Hi Elad,

Am 16.08.2016 um 10:52 schrieb Elad Ben Aharon:

Please be sure that ovirtmgmt is not part of the iSCSI bond.

 Yes, I made sure it is not part of the bond.

It does seem to have a conflict between default and enp9s0f0/ enp9s0f1.
Try to put the host in maintenance and then delete the iscsi nodes using
'iscsiadm -m node -o delete'. Then activate the host.

I tried that, I managed to get the iSCSI interface clean, no "default" anymore. But that didn't solve the problem of the host becoming "inactive". Not even the NFS domains would come up.

As soon as I remove the iSCSI-bond, the host becomes responsive again and I can activate all storage domains. Removing the bond also brings the duplicated "Iface Name" back (but this time causes no problems).

...

I wonder if there is a basic misunderstanding on my side: wouldn't it be necessary that all targets are reachable from all interfaces that are configured into the bond to make it work?

But this would either mean two interfaces in the same network or routing between the iSCSI networks.

Thanks,

Uwe
_______________________________________________
Users mailing list
Users@ovirt.org
http://lists.ovirt.org/mailman/listinfo/users