Bug 1275039

Tested with:
rhevm-3.6.1.1-0.1.el6.noarch
libvirt-1.2.17-13.el7.x86_64

I can not reproduce the bug on rhev 3.6,after connect esx 6.0,all guests can be loaded successfully without the above error.

According to the bug description,it's related to guest with iso attached,in my environment,there are several guests which has iso attached in cd.

Can you confirm whether you can use the command "virsh -c vpx://$vcenter/$datacenter/$esxi/?no_verify=1 list --all" to list all guests on your esx server?thanks.

sure I can list the vms via virsh

Refer to the error info,there is an item in $guest.vmx which libvirt can not recognize,I tried to test as below:
1.In esx 6.0,find a guest,modify it's vmx file as below:
From:
ide1:0.deviceType = "cdrom-image"
ide1:0.fileName =  "/vmfs/volumes/f0a07907-3759bde7/en_win_srv_2003_r2_enterprise_with_sp2_cd1_X13-05460.iso"                        ide1:0.present = "TRUE"

to:
ide1:0.deviceType = "cdrom-image"
ide1:0.fileName = "emptyBackingString"    ----> modify this value                                                     
ide1:0.present = "TRUE"

2.Save the above change,from vsphere client,I can see in "CD/DVD driver 1" shows iso File path as "[]",but I can not set the value as '[]' from vsphere client.

3.Use virsh to connect the esx server and dumpxml the xml of the above guest.
# virsh -c esx://$esx/?no_verify=1
Enter username for $esx [root]: 
Enter root's password for $esx: 

virsh # list --all
 Id    Name                           State
----------------------------------------------------
 1     win2012R2-vcenter              running
 33    iscsi-test                     running
 43    VMware-vcenter-server-app      running
 -     app_vcenter600                 shut off
 -     Auto-esx6.0-win7-efi           shut off
 -     bug1260689-win2008R2-GPO-Macfee shut off
 -     esx6.0-rhel5.11-i386           shut off
 -     esx6.0-rhel5.11-x86_64         shut off
 -     esx6.0-rhel6.7-i386            shut off
 -     esx6.0-rhel6.7-x86_64          shut off
 -     esx6.0-rhel7.2-x86_64          shut off
 -     esx6.0-win2003-i386            shut off
 -     esx6.0-win2003-x86_64          shut off
 -     esx6.0-win2008-i386            shut off
 -     esx6.0-win2008-x86_64          shut off


virsh # dumpxml $guest
error: internal error: Invalid or not yet handled value 'emptyBackingString' for VMX entry 'ide1:0.fileName' for device type 'cdrom-image'

4.Tried on rhev 3.6 GUI,I can load guests from esx server as steps from descrption,but the above guest is not in the guest list.

My question is since users can not set the iso file as "[]" which related to 
ide1:0.fileName = "emptyBackingString" in vmx file from vsphere client,do you have updated the vmx file for some guest on your esx server.

You can use virsh dumpxml $guest to find the guest with special configuration.

we didn't modify the guest.
I dont have that environment anymore and in ovirt we block guest that libvirt cannot parse its xml.
Libvirt should be return a legal xml for any VM it represent.

Patch proposed upstream:

https://www.redhat.com/archives/libvir-list/2016-January/msg00120.html

v2:

https://www.redhat.com/archives/libvir-list/2016-January/msg00875.html

Moving to POST:

commit 35c3aab44d20daa2623fcf2191fc2e3afc12b9f0
Author:     Michal Privoznik <mprivozn>
AuthorDate: Thu Jan 7 09:51:55 2016 +0100
Commit:     Michal Privoznik <mprivozn>
CommitDate: Mon Jan 25 08:34:23 2016 +0100

    vmx: Adapt to emptyBackingString for cdrom-image
    
    https://bugzilla.redhat.com/show_bug.cgi?id=1266088
    
    We are missing this value for cdrom-image device. It seems like
    there's no added value to extend this to other types of disk
    devices [1].
    
    1: https://www.redhat.com/archives/libvir-list/2016-January/msg01038.html
    
    Signed-off-by: Michal Privoznik <mprivozn>

on libvirt version: 
===================
libvirt-daemon-1.2.17-13.el7_2.3.x86_64

its faling the following if statement:
======================================

for vm in conn.listAllDomains():
    params['vmName'] = vm.name()
    if vm.state()[0] == libvirt.VIR_DOMAIN_SHUTOFF:
        params['status'] = "Down"
    else:
        params['status'] = "Up"

with the exception:
===================
Traceback (most recent call last):
  File "/usr/share/vdsm/v2v.py", line 156, in get_external_vms
    _add_vm_info(vm, params)
  File "/usr/share/vdsm/v2v.py", line 662, in _add_vm_info
    if vm.state()[0] == libvirt.VIR_DOMAIN_SHUTOFF:
  File "/usr/lib64/python2.7/site-packages/libvirt.py", line 2661, in state
    if ret is None: raise libvirtError ('virDomainGetState() failed', dom=self)
libvirtError: Domain not found: Could not find domain with UUID '4216df1f-d02a-adc9-d0c8-e2c9e9ab2ef2'


when I comment the if with statement I get error on the following code:
======================================================================
xml = vm.XMLDesc(0)

(In reply to Shahar Havivi from comment #15)
> on libvirt version: 
> ===================
> libvirt-daemon-1.2.17-13.el7_2.3.x86_64
> 
> its faling the following if statement:
> ======================================
> 
> for vm in conn.listAllDomains():
>     params['vmName'] = vm.name()
>     if vm.state()[0] == libvirt.VIR_DOMAIN_SHUTOFF:
>         params['status'] = "Down"
>     else:
>         params['status'] = "Up"
> 
> with the exception:
> ===================
> Traceback (most recent call last):
>   File "/usr/share/vdsm/v2v.py", line 156, in get_external_vms
>     _add_vm_info(vm, params)
>   File "/usr/share/vdsm/v2v.py", line 662, in _add_vm_info
>     if vm.state()[0] == libvirt.VIR_DOMAIN_SHUTOFF:
>   File "/usr/lib64/python2.7/site-packages/libvirt.py", line 2661, in state
>     if ret is None: raise libvirtError ('virDomainGetState() failed',
> dom=self)
> libvirtError: Domain not found: Could not find domain with UUID
> '4216df1f-d02a-adc9-d0c8-e2c9e9ab2ef2'
> 
> 
> when I comment the if with statement I get error on the following code:
> ======================================================================
> xml = vm.XMLDesc(0)


Is it the same issue with bug 1287681?

(In reply to tingting zheng from comment #16)
> (In reply to Shahar Havivi from comment #15)
> > 
> > when I comment the if with statement I get error on the following code:
> > ======================================================================
> > xml = vm.XMLDesc(0)
> 
> 
> Is it the same issue with bug 1287681?

I'm afraid so. if you try another domain does it work?

It fail for some of the domains and some works fine

(In reply to Michal Privoznik from comment #17)
> (In reply to tingting zheng from comment #16)
> > (In reply to Shahar Havivi from comment #15)
> > > 
> > > when I comment the if with statement I get error on the following code:
> > > ======================================================================
> > > xml = vm.XMLDesc(0)
> > 
> > 
> > Is it the same issue with bug 1287681?
> 
> I'm afraid so. if you try another domain does it work?

check bug 1287681 comment 13, # virsh list --all works some times  and fails a few seconds later.

We have no problem via virsh lists --all, its working good.
The problem is the python API as described in comment #15

based on comment 20, would you like us to reopen this bug or open a new bug?

I don't know if that is a new bug...
I do think its the same - Nisim said that he can share his environment if its needed.

(In reply to Nisim Simsolo from comment #23)
> It looks like vSphere 5.5 bug:
> I've succeeded to create similar VM with empty datastore ISO file on TLV
> VMware environment (Version 6.0.0 Build 2594327), and when "loading" VMs
> using webadmin import dialog, VMware VMs are listed except the problematic
> VM (FAD).
> 
> bug: https://bugzilla.redhat.com/show_bug.cgi?id=1287681 is mentioning this
> issue. 
> 
> If required, I can open a bug to VMware related to this issue.

That would be great! I would have done it myself but as I said in the linked bug I have no idea where to report that.

I can reproduce it with:

libvirt-1.3.1-1.el7.x86_64

1,Modify vmx file on the esx server
From:
ide1:0.deviceType = "cdrom-image"
ide1:0.fileName = "/vmfs/volumes/f0a07907-3759bde7/en_win_srv_2003_r2_enterprise_with_sp2_cd1_X13-05460.iso" ide1:0.present = "TRUE"

to:
ide1:0.deviceType = "cdrom-image"
ide1:0.fileName = "emptyBackingString" ----> modify this value
ide1:0.present = "TRUE"

2, Use virsh dumpxml the xml of the above guest

    virsh -c vpx://root.145.47/data/10.66.144.40/?no_verify=1 dumpxml empty_cdrom
    Enter root's password for 10.66.145.47:
    error: internal error: Invalid or not yet handled value 'emptyBackingString' for VMX entry 'ide0:0.fileName' for device type 'cdrom-image'

3,Go to rhevm GUI , the above guest can not be loaded

Test it with

Libvirt-1.3.3-2.el7.x86_64
virt-v2v-1.32.3-1.el7.x86_64


1,After step 2:

    virsh -c vpx://root.145.47/data/10.66.144.40/?no_verify=1 dumpxml empty_cdrom
    Enter root's password for 10.66.145.47:
    ...
    <devices>
    <disk type='file' device='disk'>
    <source file='[datastore1] empty_cdrom/empty_cdrom.vmdk'/>
    <target dev='sda' bus='scsi'/>
    <address type='drive' controller='0' bus='0' target='0' unit='0'/>
    </disk>
    <disk type='file' device='cdrom'>
    <target dev='hda' bus='ide'/>
    <address type='drive' controller='0' bus='0' target='0' unit='0'/>
    </disk>
    ...
    2,Go to rhevm ,the above guest can be loaded successfully
    3,Using the latest version virt-v2v convert it to kvm,it's can be boot normally 
    As the above comments ,the bug has already been fixed .So move it to VERIFIED

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://rhn.redhat.com/errata/RHSA-2016-2577.html