Open Stack

Hi,

unfortunately, I don't have an answer for you, but it seems that  
you're not alone with this. In the past 10 days or so I have read  
about very similiar issues multiple times (e.g. [1], [2]). In fact, it  
sounds like the update could be responsible for these changes.

Usually, you can change the disk_bus by specifying glance image  
properties, something like this:

openstack image set --property hw_scsi_model=virtio-scsi --property  
hw_disk_bus=scsi --property hw_qemu_guest_agent=yes --property  
os_require_quiesce=yes <IMAGE_ID>

But I doubt any effect of this, there has to be something else telling  
libvirt to use scsi instead of virtio. I hope someone else has an idea  
where to look at since I don't have this issue and can't reproduce it.

What is your output for

---cut here---
root at compute:~ # grep -A3 virtio-blk  
/usr/lib/udev/rules.d/60-persistent-storage.rules
# virtio-blk
KERNEL=="vd*[!0-9]", ATTRS{serial}=="?*",  
ENV{ID_SERIAL}="$attr{serial}",  
SYMLINK+="disk/by-id/virtio-$env{ID_SERIAL}"
KERNEL=="vd*[0-9]", ATTRS{serial}=="?*",  
ENV{ID_SERIAL}="$attr{serial}",  
SYMLINK+="disk/by-id/virtio-$env{ID_SERIAL}-part%n"
---cut here---

You could also take a look into  
/etc/glance/metadefs/compute-libvirt-image.json, maybe there is  
something wrong there, but as I said, I can't really reproduce this.

Good luck!

[1]  
https://ask.openstack.org/en/question/112488/libvirt-not-allocating-cpu-and-disk-to-vms-after-the-os-update/
[2] https://bugs.launchpad.net/nova/+bug/1560965


Zitat von Yedhu Sastry <yedhusastri at gmail.com>:

> Hello,
>
> I have an OpenStack cluster(Newton) which is basically a test cluster.
> After the regular OS security update and upgrade in all my compute nodes I
> have problem with New VMs. While launching new VM's Iam getting the
> Error  "ALERT!
> LABEL=cloudimg-rootfs does not exist  Dropping to a shell!" in the console
> log of VM's. In horizon it is showing as active. Iam booting from image not
> from volume. Before the update everything was fine.
>
> Then I checked all the logs related to OpenStack and I cant find any info
> related to this. I spent days and I found that after the update libvirt is
> now using scsi instead of virtio. I dont know why. All the VM's which I
> created before the update are running fine and  is using 'virtio'. Then I
> tried to manually change the instancexx.xml file of the libvirt to use "
> <target dev='vda' bus='virtio'/> " and started the VM again using 'virsh
> start instancexx'. VM got started and then went to shutdown state. But in
> the console log I can see VM is getting IP and properly booting without any
> error and then it goes to poweroff state.
>
>
> 1) Whether this issue is related to the update of libvirt?? If so why
> libvirt is not using virtio_blk anymore?? Why it is using only
> virtio_scsi?? Is it possible to change libvirt to use virtio_blk instead of
> virtio_scsi??
>
> 2) I found nova package version on compute nodes are 14.0.10 and on
> controller node it is 14.0.1. Whether this is the cause of the problem??
> Whether an update in controller node solve this issue?? Iam not sure about
> this.
>
> 3) Why Task status of  instancexx is showing as Powering Off in horizon
> after 'virsh start instancexx' in the compute node?? Why it is not starting
> the VM with the manually customized .xml file of libvirt??
>
>
> Any help is really appreciated.
>
>
> --
>
> Thank you for your time and have a nice day,
>
>
> With kind regards,
> Yedhu Sastri



-- 
Eugen Block                             voice   : +49-40-559 51 75
NDE Netzdesign und -entwicklung AG      fax     : +49-40-559 51 77
Postfach 61 03 15
D-22423 Hamburg                         e-mail  : eblock at nde.ag

         Vorsitzende des Aufsichtsrates: Angelika Mozdzen
           Sitz und Registergericht: Hamburg, HRB 90934
                   Vorstand: Jens-U. Mozdzen
                    USt-IdNr. DE 814 013 983