Newly deployed ovirt instance can not start a VM
Hello everyone! I just finished setting up our new demo Ovirt implementation, but am having some issues once I get to the point of starting up a VM. Every time we try to start a VM, we get the error: The host {hostname} did not satisfy internal filter Network because network(s) are missing. This is trying to launch vm with the Cirros image from the OVirt glance repository. This is confusing, because we only have one network defined (the "ovirtmgmt" network), and each of our 4 nodes shows it attached and indicated as functional. Our nodes are all CentOS 6.6 installations, with the ovirtmgmt bridge bound to eth2. If I go to Networks/ovirtmgmt under my datacenter, and click on the Networks tab at the top, select the ovirtmgmt network, and Hosts tab at the bottom, it shows that ovirtmgmt network is up on all 4 nodes. Even if I tell the VM to not attach itself to any networks at all, the machine fails to start with the same error. Have done a lot of google searching to no avail. We're also worried we did not get our gluster storage network set up right. We added the Storage Domain as type Data (Master)/GlusterFS, but it looks like Ovirt is using Fuse rather than the new libgfapi functionality - since we can see the gluster mounted /rhev/data-center/mnt/glusterSD/ with fuse.glusterfs, which would indicate we are not using the native Gluster integration. However, the latest info I can find seems to indicate that the gluster integration should be working in latest CentOS builds. Thanks for any help anyone can give me! The boss wants this demo sooner rather than later!
Jeremy Utley <jeremy@ifuzioncorp.com> schreef op 6 december 2014 00:21:22 CET:
Hello everyone!
I just finished setting up our new demo Ovirt implementation, but am having some issues once I get to the point of starting up a VM. Every time we try to start a VM, we get the error:
The host {hostname} did not satisfy internal filter Network because network(s) are missing.
Try looking at vdsm.log on the host where the vm is being started. You can force a host by using runonce and the host tab. Looking at engine.log migt also help
This is trying to launch vm with the Cirros image from the OVirt glance repository. This is confusing, because we only have one network defined (the "ovirtmgmt" network), and each of our 4 nodes shows it attached and indicated as functional.
Our nodes are all CentOS 6.6 installations, with the ovirtmgmt bridge bound to eth2. If I go to Networks/ovirtmgmt under my datacenter, and click on the Networks tab at the top, select the ovirtmgmt network, and Hosts tab at the bottom, it shows that ovirtmgmt network is up on all 4 nodes. Even if I tell the VM to not attach itself to any networks at all, the machine fails to start with the same error. Have done a lot of google searching to no avail.
We're also worried we did not get our gluster storage network set up right. We added the Storage Domain as type Data (Master)/GlusterFS, but it looks like Ovirt is using Fuse rather than the new libgfapi functionality - since we can see the gluster mounted /rhev/data-center/mnt/glusterSD/ with fuse.glusterfs, which would indicate we are not using the native Gluster integration. However, the latest info I can find seems to indicate that the gluster integration should be working in latest CentOS builds. It should start working as soon as a patch is in mainline. It doesn't right now but it will if you use glusterfs as a data store. There is a recent thread on the ML about it with me as OP.
Joop
On 06/12/14 01:21, Jeremy Utley wrote:
Hello everyone!
I just finished setting up our new demo Ovirt implementation, but am having some issues once I get to the point of starting up a VM. Every time we try to start a VM, we get the error:
The host {hostname} did not satisfy internal filter Network because network(s) are missing.
This reminds me of issues with missing boot protocol for the display network on hosts (there's a bug on improving the error message for that). Does ovirtmgmt have a boot protocol and IP address configured on the hosts?
This is trying to launch vm with the Cirros image from the OVirt glance repository. This is confusing, because we only have one network defined (the "ovirtmgmt" network), and each of our 4 nodes shows it attached and indicated as functional.
Our nodes are all CentOS 6.6 installations, with the ovirtmgmt bridge bound to eth2. If I go to Networks/ovirtmgmt under my datacenter, and click on the Networks tab at the top, select the ovirtmgmt network, and Hosts tab at the bottom, it shows that ovirtmgmt network is up on all 4 nodes. Even if I tell the VM to not attach itself to any networks at all, the machine fails to start with the same error. Have done a lot of google searching to no avail.
We're also worried we did not get our gluster storage network set up right. We added the Storage Domain as type Data (Master)/GlusterFS, but it looks like Ovirt is using Fuse rather than the new libgfapi functionality - since we can see the gluster mounted /rhev/data-center/mnt/glusterSD/ with fuse.glusterfs, which would indicate we are not using the native Gluster integration. However, the latest info I can find seems to indicate that the gluster integration should be working in latest CentOS builds.
Thanks for any help anyone can give me! The boss wants this demo sooner rather than later!
_______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users
participants (3)
-
Jeremy Utley -
Joop van de Wege -
Lior Vernia