Wind River Support Network

HomeDefectsOVP-1997
Fixed

OVP-1997 : OVP 6 Problems With Spice/oVirt

Created: Jul 21, 2014    Updated: Mar 11, 2016
Resolved Date: Aug 28, 2014
Found In Version: 5.0.1.15,6.0
Fix Version: 6.0.0.11
Severity: Standard
Applicable for: Wind River Linux 5
Component/s: Documentation

Description

We should modify the docs and Twiki pages to steer everyone away from spice and towards VNC unless their host is Ubuntu 14.04 or the latest(?) Fedora.
Spice does not work with Windows or Ubuntu 12.04 hosts/browsers, and the only place this is clearly mentioned is on http://twiki.wrs.com/PBUeng/LinuxOVP6UseCases 

Also, the OVP Virtual Node Users Guide is unhelpful in regards to spice (page 152):
1.	Does not make it clear that Ubuntu 12.04 and Windows hosts are unsupported for spice. So you spend a lot of time creating a VM that you can’t use.
2.	Does not tell you how to connect (to what, the node or the VM?) to the VM with spice or VNC.   [You are connecting to the ovirt-node.]
3.	Does not tell you what the IP address or port of the node or the VM is.   [It is on the ovirt-node’s ovirtmgmt interface, which in my case is the Operations And Maintenance (OAM) network and the building public network.]
4.	Is probably out of sequence: can you really connect spice or VNC to a VM where the guest image has not even been installed yet.   [Not an issue, since you are connecting to the ovirt-node.]
5.	Hard to tell yet whether the guest installation procedure on page 38 will work (does not look promising).   [They did work, except for the new issue #10 below]

I am going to have to delete vm_1 and go back to Rob Woolley’s OVP evaluation manual (README_OVP_EVAL_V2.pdf – attached) to try to bring the VM up with VNC. See section 2.14 of attached README_OVP_EVAL_V2.  [I was able to shut down the VM and change the console to VNC. Later I was able to shut down the hard disk and enlarge it by 3GB. In both cases I did not have to start over.]

When booted, the VM got a valid IP address on the OAM network, which is convenient.

Virtual Node User’s Guide Documentation Errata
1.	Doesn’t tell you how to handle auto-registered ovirt-node (workflow, setting root password, etc.), although it advises the user to auto-register ovirt-node.
2.	Page 136 screen shot missing colon (:) from ovirt-node.wrs.com:/export/nfs/data. Also these operations take a lot longer than “a few seconds”. “A minute” is much closer.
3.	Pages 143-148 “Operating System” should ALWAYS be “Linux”, not “Other OS”. It is “Linux” on the first screen shot.
4.	Various screen shots and work flows are not quite correct and need updates.
5.	Page 149, step 7: no need to create nic1 – it was auto-created. But where did the MAC ID come from?
6.	Page 150, step 8: “Red Hat VirtIO” does not exist, “VirtIO” must be selected, I suppose.
7.	Page 150-step 9: What to do with the “ACTIVATE” checkbox? It is checked by default – is that correct?
8.	Spice issues mentioned above. Need to make VNC the default for most users.
9.	On page 150 step 9, there is a comment: “NOTE: The OVP guest ext3 image is less than 1 GB in size because it is an initramfs system. Use a size that is appropriate for your image.” This led me to try a 2GB disk size, which ran out of space and made the installation fail. I tried again with a 5GB disk size and the installation succeeded. So the comment (if not deleted) should probably include a multiplier (5?) on the size of the ext3 or iso file.
10.	The guest -> VM installation process (page 40) does not have all the prompts indicated and does not have a text console to issue the reboot command. It gets caught in an infinite re-install loop if you click the “Reboot” button at the end of the installation, because it tries to boot again off the installer “CD-ROM” (iso file) instead of the new vda hard disk. The solution is: When presented with the “Reboot” button, power down the VM and then start it up again (it will then boot off its vda hard drive instead of the installer “CD-ROM”).
11. It's not clear to me whether the Linux boot command line (used by Grub) must have "selinux=0", or whether "enforcing=0" is sufficient. See the note on http://twiki.wrs.com/PBUeng/LinuxOVP6UseCases where it says "Make sure to disable SELinux. ". Is this on the development host, node, or engine? What is the correct kernel command line setting for "enforcing=?" and "selinux=?" on each of these systems?
Live chat
Online