I just installed a new 3.1 engine and I have one node with the latest node iso installed. After setting up one of the network interfaces on the node and confirming I have network I am able to attach it to the engine. I've done this starting from both the node and the engine and it seems to work either way. Everything looks good until I go to the network controls in the web interface. I need to configure another network interface but to do so the ovirtmgmt network must be assigned to an interface. I'm not sure why it isn't already since the node is already assigned to a cluster. If I attempt to assign the management network it starts reconfiguring the node and then irrecoverably breaks the network on the node. By irrecoverable I mean I have to manually log in to the node as root and use ifdown to shut off all of the interfaces, delete all of the network configs under /config and then restart the machine and re-setup the network to get it back. Nothing less seems to !

tore network to the node. From what I can tell when the node is assigned to a cluster the network looks good on the node side. It's running a bridge named ovirtmgmt and seems to be happy. When the engine tries to assign ovirtmgmt to the node it creates a new bridge named after the network interface but is unable to start it. I've done a lot troubleshooting of network issues on 3.0 and have been able to work around them, but I seem to be at an impasse here. Without the additional network config my VM's won't work. Is there another way to force the network configuration without using the web interface? I know what the network setup should be. I've done all kinds of manual bridging, bonding, etc. under Fedora. I just can't seem to get oVirt to understand. -Jacob

_______________________________________________ Users mailing list Users(a)ovirt.org http://lists.ovirt.org/mailman/listinfo/users

OK, so I reinstalled yesterday using the prepackaged 3.1 beta yum repository. It's better. It now correctly assigns the ovirtmgmt network to the node when it is attached.

Unfortunately, I still cannot seem to add a second interface or VLAN without breaking the networking again.

-Jacob

________________________________________ From: Livnat Peer [lpeer(a)redhat.com] Sent: Sunday, August 05, 2012 4:59 AM To: Jacob Wyatt Cc: users(a)oVirt.org Subject: Re: [Users] 3.1 network setup On 04/08/12 17:51, Jacob Wyatt wrote: > I just installed a new 3.1 engine and I have one node with the latest node iso installed. After setting up one of the network interfaces on the node and confirming I have network I am able to attach it to the engine. I've done this starting from both the node and the engine and it seems to work either way. Everything looks good until I go to the network controls in the web interface. I need to configure another network interface but to do so the ovirtmgmt network must be assigned to an interface. I'm not sure why it isn't already since the node is already assigned to a cluster. If I attempt to assign the management network it starts reconfiguring the node and then irrecoverably breaks the network on the node. By irrecoverable I mean I have to manually log in to the node as root and use ifdown to shut off all of the interfaces, delete all of the network configs under /config and then restart the machine and re-setup the network to get it back. Nothing less seems to!

res > tore network to the node. From what I can tell when the node is assigned to a cluster the network looks good on the node side. It's running a bridge named ovirtmgmt and seems to be happy. When the engine tries to assign ovirtmgmt to the node it creates a new bridge named after the network interface but is unable to start it. I've done a lot troubleshooting of network issues on 3.0 and have been able to work around them, but I seem to be at an impasse here. Without the additional network config my VM's won't work. Is there another way to force the network configuration without using the web interface? I know what the network setup should be. I've done all kinds of manual bridging, bonding, etc. under Fedora. I just can't seem to get oVirt to understand. > > -Jacob Hi Jacob, You installed the engine from source? if so what is the engine hash commit you used for installation? The reason I am asking that is because we had a major regression last week were we broke the network API. It was fixed in commit fa95ed435ac824ae7fc721239d862337053202bc, I suggest you don't cherry pick this specific commit but rebase on it. Thanks, Livnat > > _______________________________________________ > Users mailing list > Users(a)ovirt.org > http://lists.ovirt.org/mailman/listinfo/users > _______________________________________________ Users mailing list Users(a)ovirt.org http://lists.ovirt.org/mailman/listinfo/users

I'm using the iso images for oVirt Node Hypervisor release 2.5.0 (2.0.fc17). I was able to manually edit the ifcfg files manually run the persist command against them, reboot, and the engine sees everything just fine. The engine was able to successfully configure a VLAN interface and then add ovirtmgmt as a bridge. The problem was adding a second vlan. Instead of just creating a second VLAN interface and bridge it tried to configure a bridge directly on the network interface. I'll look at it in more detail when I have time. Right now I'm having trouble actually starting a VM so I need to work that out so I can actually test the networking.

________________________________________ From: Livnat Peer [lpeer(a)redhat.com] Sent: Monday, August 06, 2012 2:04 AM To: Jacob Wyatt Cc: users(a)oVirt.org Subject: Re: [Users] 3.1 network setup On 06/08/12 05:06, Jacob Wyatt wrote: > OK, so I reinstalled yesterday using the prepackaged 3.1 beta yum repository. It's better. It now correctly assigns the ovirtmgmt network to the node when it is attached. > Unfortunately, I still cannot seem to add a second interface or VLAN without breaking the networking again. I'm wondering if part of it is an issue with the node. If I try to configure networking through the admin interface on the node it only lets me configure one interface. Configuring a second interface breaks the first one. I may try manually editing the config files on the node, but I'm not sure how the engine will see that. > > -Jacob > manually editing the node network should work just fine, you should put your node in maintenance mode via the UI, do the manual changes you want and then activate the host which should reflect the new host configuration. If there is an issue with the above I'll be happy if you can share it with us so we can fix that. Having said the above It would really help us if you can avoid using the manual edit and let us find what is wrong with the flow from the UI. What vdsm version is installed on your host? did you build from source? Is the second network you are trying to add is a marked as a vm-network? (vm-network is later translated to bridge on the node) we block more than one non-vm network on an interface. can you send the logs (vdsm + engine)? Livnat > ________________________________________ > From: Livnat Peer [lpeer(a)redhat.com] > Sent: Sunday, August 05, 2012 4:59 AM > To: Jacob Wyatt > Cc: users(a)oVirt.org > Subject: Re: [Users] 3.1 network setup > > On 04/08/12 17:51, Jacob Wyatt wrote: >> I just installed a new 3.1 engine and I have one node with the latest node iso installed. After setting up one of the network interfaces on the node and confirming I have network I am able to attach it to the engine. I've done this starting from both the node and the engine and it seems to work either way. Everything looks good until I go to the network controls in the web interface. I need to configure another network interface but to do so the ovirtmgmt network must be assigned to an interface. I'm not sure why it isn't already since the node is already assigned to a cluster. If I attempt to assign the management network it starts reconfiguring the node and then irrecoverably breaks the network on the node. By irrecoverable I mean I have to manually log in to the node as root and use ifdown to shut off all of the interfaces, delete all of the network configs under /config and then restart the machine and re-setup the network to get it back. Nothing less seems t!

! > res >> tore network to the node. From what I can tell when the node is assigned to a cluster the network looks good on the node side. It's running a bridge named ovirtmgmt and seems to be happy. When the engine tries to assign ovirtmgmt to the node it creates a new bridge named after the network interface but is unable to start it. I've done a lot troubleshooting of network issues on 3.0 and have been able to work around them, but I seem to be at an impasse here. Without the additional network config my VM's won't work. Is there another way to force the network configuration without using the web interface? I know what the network setup should be. I've done all kinds of manual bridging, bonding, etc. under Fedora. I just can't seem to get oVirt to understand. >> >> -Jacob > > > Hi Jacob, > > You installed the engine from source? if so what is the engine hash > commit you used for installation? > > The reason I am asking that is because we had a major regression last > week were we broke the network API. It was fixed in commit > fa95ed435ac824ae7fc721239d862337053202bc, I suggest you don't cherry > pick this specific commit but rebase on it. > > Thanks, Livnat > > > >> >> _______________________________________________ >> Users mailing list >> Users(a)ovirt.org >> http://lists.ovirt.org/mailman/listinfo/users >> > > > > _______________________________________________ > Users mailing list > Users(a)ovirt.org > http://lists.ovirt.org/mailman/listinfo/users > _______________________________________________ Users mailing list Users(a)ovirt.org http://lists.ovirt.org/mailman/listinfo/users