libvirt users - Sep 2013 - Error Attaching Seventh VirtIO-SCSI Device to Guest

I have run into a problem attempting to attach the seventh virtio-scsi device to
a RHEL 6.4 Guest from a RHEL 6.4 host running libvirt version 0.10.2-18.
I have a guest that is running RHEL6.4 where I can attach disks sda(boot), sdb,
sdc, sdd, sde and sdf but when I try to attach sdg the virsh attach-disk command
fails with the error:
error: Failed to attach disk
error: internal error Unable to determine model for scsi controller

I logged Bug 996327  on this issue including an attached script to repro the bug
but there has been no activity on that bug so far.

This is the results when attaching virtio-scsi disks in a loop. Sdf is attached
but sdg fails.

virsh attach-disk --domain UEM004n1 --source /dev/GuestVols/UEM004-5 --target
sdf --cache none --type disk --config --wwn 1fc7058865439717 --serial
c7d11fc7058865439717 --shareable
Disk attached successfully

Vol UEM004-6 created
virsh attach-disk --domain UEM004n1 --source /dev/GuestVols/UEM004-6 --target
sdg --cache none --type disk --config --wwn 4d3ddb9d15b05e7e --serial
62fd4d3ddb9d15b05e7e --shareable

error: Failed to attach disk
error: internal error Unable to determine model for scsi controller


The script to create and attach LVM volumes using virtio-scsi is available here:
https://bugzilla.redhat.com/attachment.cgi?id=785925

Has anyone else run into this... I had the understanding that each
virtio-controller could support 4K devices.

  --jim

On 04/09/13 09:34, McEvoy, James wrote:
> I have run into a problem attempting to attach the seventh virtio-scsi
device to a RHEL 6.4 Guest from a RHEL 6.4 host running libvirt version
0.10.2-18.
> I have a guest that is running RHEL6.4 where I can attach disks sda(boot),
sdb, sdc, sdd, sde and sdf but when I try to attach sdg the virsh attach-disk
command fails with the error:
> error: Failed to attach disk
> error: internal error Unable to determine model for scsi controller
>
> I logged Bug 996327  on this issue including an attached script to repro
the bug but there has been no activity on that bug so far.
I took the bug.
>
> This is the results when attaching virtio-scsi disks in a loop. Sdf is
attached but sdg fails.
>
> virsh attach-disk --domain UEM004n1 --source /dev/GuestVols/UEM004-5
--target sdf --cache none --type disk --config --wwn 1fc7058865439717 --serial
c7d11fc7058865439717 --shareable
> Disk attached successfully
>
> Vol UEM004-6 created
> virsh attach-disk --domain UEM004n1 --source /dev/GuestVols/UEM004-6
--target sdg --cache none --type disk --config --wwn 4d3ddb9d15b05e7e --serial
62fd4d3ddb9d15b05e7e --shareable
>
> error: Failed to attach disk
> error: internal error Unable to determine model for scsi controller
It looks odd,  It looks to me there is new scsi controller is generated
automatically by libvirt when the 6th volume is attaching to the domain,
since for narrow SCSI bus, we allow 7 unit for each SCSI controller. But
it's strange that it couldn't get a default model for the new SCSI 
controller.

Can you get the devices your qemu supports by:

     $ qemu-kvm -device ?

And attach your domain conf?

Either in the list or in the bug is fine.

Osier

> -----Original Message-----
> From: Osier Yang [mailto:jyang@redhat.com]
> Sent: Friday, September 06, 2013 10:54 PM
> To: McEvoy, James
> Cc: libvirt-users@redhat.com
> Subject: Re: [libvirt-users] Error Attaching Seventh VirtIO-SCSI Device to
Guest
> 
> On 04/09/13 09:34, McEvoy, James wrote:
> > I have run into a problem attempting to attach the seventh virtio-scsi
device
> to a RHEL 6.4 Guest from a RHEL 6.4 host running libvirt version 0.10.2-18.
> > I have a guest that is running RHEL6.4 where I can attach disks
sda(boot), sdb,
> sdc, sdd, sde and sdf but when I try to attach sdg the virsh attach-disk
> command fails with the error:
> > error: Failed to attach disk
> > error: internal error Unable to determine model for scsi controller
> >
> > I logged Bug 996327  on this issue including an attached script to
repro the
> bug but there has been no activity on that bug so far.
> 
> I took the bug.
> 
> >
> > This is the results when attaching virtio-scsi disks in a loop. Sdf is
attached
> but sdg fails.
> >
> > virsh attach-disk --domain UEM004n1 --source /dev/GuestVols/UEM004-5
> > --target sdf --cache none --type disk --config --wwn 1fc7058865439717
> > --serial c7d11fc7058865439717 --shareable Disk attached successfully
> >
> > Vol UEM004-6 created
> > virsh attach-disk --domain UEM004n1 --source /dev/GuestVols/UEM004-6
> > --target sdg --cache none --type disk --config --wwn 4d3ddb9d15b05e7e
> > --serial 62fd4d3ddb9d15b05e7e --shareable
> >
> > error: Failed to attach disk
> > error: internal error Unable to determine model for scsi controller
> 
> It looks odd,  It looks to me there is new scsi controller is generated
> automatically by libvirt when the 6th volume is attaching to the domain,
since
> for narrow SCSI bus, we allow 7 unit for each SCSI controller. But it's
strange
> that it couldn't get a default model for the new SCSI controller.
> 
> Can you get the devices your qemu supports by:
> 
>      $ qemu-kvm -device ?
> 
> And attach your domain conf?
> 
> Either in the list or in the bug is fine.
> 
> Osier
Hi Osier,

I have attached the details you asked for to the bug 996327 and here is the
short version:

These are the qemu-kvm -drive and -device options for /dev/sdg on the guest:
-drive
file=/dev/GuestVols/UEM002-6,if=none,id=drive-scsi0-0-0-6,format=raw,serial=8aad44f9e53e05868e81,cache=none
-device
scsi-hd,bus=scsi0.0,channel=0,scsi-id=0,lun=6,drive=drive-scsi0-0-0-6,id=scsi0-0-0-6,wwn=44f9e53e05868e81

I am seeing an error on device /dev/sdg on the guest when I query SCSI page 0x83
for that device.
[root@node1 ~]# sg_inq -p 0x83 /dev/sdg
VPD INQUIRY: Device Identification page
  Designation descriptor number 1, descriptor length: 24
    designator_type: vendor specific [0x0],  code_set: ASCII
    associated with the addressed logical unit
      vendor specific: 22b0daf285d88050fd39
  Designation descriptor number 2, descriptor length: 12
    designator_type: NAA,  code_set: Binary
    associated with the addressed logical unit
      << unexpected naa [0xd]>>
 00     da f2 85 d8 80 50 fd 39                             .....P.9        
[root@node1 ~]#

One other detail, I am defining the scsi controller using the virt-install
command.

  --jim