headers
Bryan Kearney | 1 Dec 14:32
Picon
Favicon
Gravatar

Re: Making a disk image created with virt-install qcow2 format from the start

Stephan wrote:
> Hi,
> 
> I've been creating images with virt-install (thanks by the way!)
> 
> I was just wondering, is there a way to specify the disk image file type
> when you create the machine? I can convert it using tools from qemu, but
> it's an extra step and I'd prefer to have the images as qcow2 from the
> start if possible.
> 
> I'm doing this:
> 
> sudo mkdir /vservers/edi1-whatever1/ && sudo virt-install --hvm
> --accelerate --os-type=linux --os-variant=debianEtch --arch=x86_64
> --bridge=br0 --ram=256 --vcpus=4 --check-cpu --vnc --name=edi1-whatever1
> --file=/vservers/edi1-whatever1/disk0.qcow2 --file-size=8
> --cdrom=/dev/cdrom --debug
> 
> but the file isn't the right format:
> 
> file disk0.qcow2 
> disk0.qcow2: x86 boot sector; partition 1: ID=0x83, active, starthead 1,
> startsector 63, 15952482 sectors; partition 2: ID=0x5, starthead 0,
> startsector 15952545, 819315 sectors
> 
> instead of something like this:
> 
> file disk0.qcow2 
> disk0.qcow2: Qemu Image, Format: Qcow , Version: 2
>
(Continue reading)

Permalink | Reply | 0 comments |
headers
Cole Robinson | 1 Dec 15:40
Picon
Favicon

Re: Réf. : Re: [et-mgmt-tools] virt-manager on windows workstation ?

FCOMBERNOUS@... wrote:
> -----Cole Robinson <crobinso@...> a écrit : -----
>> FCOMBERNOUS@... wrote:
>>>  Lo,
>>>
>>>  Is it possible to run virt-manager on windows workstation ?
>>>
>>>  I hope my google search is wrong. In a short way, it says "No".
>>>
>>> In this case, virt-manager would be a "simple" acces point to the
>>> demon libvirt running on a linux host.
>> No one has done it yet, but it should be possible. virt-manager would
>> need patches though.
> 
> So, python and something like cygwin, installed on the workstation is not
> enough ?
> 

That could be sufficient, but I'm really not sure what would be
required. If you give it a shot, I'd be interested in what steps it
takes or bugs you hit.

Thanks,
Cole
Permalink | Reply | 0 comments |
headers
Cole Robinson | 1 Dec 15:43
Picon
Favicon

Re: Connecting to a remote libvirtd host using virt-manger

Jon Smith wrote:
> virt-manager and libvirtd both functioning locally.  I tried to use
> virt-manager to access the same libvirtd host remotely but I'm having
> some trouble.
> 
> I'm using qemu+ssh to connect to the remote host, but when I do,
> virt-manager turns gray and becomes unresponsive, eventually the part
> of the window showing the hosts goes blank.  If I stop libvirtd on the
> remote host, virt-manager refreshes itself and shows the libvirtd host
> as connected, and even lists all the domains available.  I'm also able
> to connect using virsh and that works fine.
> 
> I can repeat this process over and over again, even after rebooting
> the hosts involved.  I've also run libvirtd with the --verbose flag
> and can send the output (~80K) if that would help.
> 

What distro are you using? What virt-manager version? Distro package or
compiled from source?

How are you invoking virt-manager? Can you try running 'virt-manager
--no-fork' from the command line, trying to connect to the remote host,
and seeing if a password prompt pops up in the terminal?

Thanks,
Cole
Permalink | Reply | 2 comments |
headers
Cole Robinson | 1 Dec 16:07
Picon
Favicon

Re: [PATCH] Fix lookup pool by path

Andreas Schneider wrote:
> On Tuesday 25 November 2008 16:08:25 Andreas Schneider wrote:
>> Hi,
>>
>> attached is a patch for virtinst to fix the lookup_pool_by_path() function.
>> It removes the trailing slash to have a normalized path.
>>
> 
> I've updated the patch. It works now :)
> 
> 

Hmm, so your pool xml lists the target directory with a trailing slash?
I'm not seeing that. What version of libvirt are you using, and can you
provide the pools xml?

Thanks,
Cole
Permalink | Reply | 4 comments |
headers
Cole Robinson | 1 Dec 16:12
Picon
Favicon

Re: [PATCH] Use netcat if found

Andreas Schneider wrote:
> Hi,
> 
> attached is a patch for virt-manager to use the netcat binary instead of nc if 
> it is found.
> 
> 
> Cheers,
> 
> 
> 	-- andreas
> 

I'd rather keep things operating the same if 'nc' is present, which
would mean using 'netcat' only if 'nc' isn't available. Is there a
reason we should prefer netcat over nc, or is it just that your distro
provides one and not the other?

Thanks,
Cole
Permalink | Reply | 2 comments |
headers
Daniel P. Berrange | 1 Dec 16:26
Picon
Favicon

Re: [PATCH] Use netcat if found

On Mon, Dec 01, 2008 at 10:12:52AM -0500, Cole Robinson wrote:
> Andreas Schneider wrote:
> > Hi,
> > 
> > attached is a patch for virt-manager to use the netcat binary instead of nc if 
> > it is found.
> > 
> > 
> > Cheers,
> > 
> > 
> > 	-- andreas
> > 
> 
> I'd rather keep things operating the same if 'nc' is present, which
> would mean using 'netcat' only if 'nc' isn't available. Is there a
> reason we should prefer netcat over nc, or is it just that your distro
> provides one and not the other?

This used to be the case with Debian/Ubuntu, but they now provide the
'nc' binary too. In addition the 'netcat' binary they also ship does
not work with libvirt since it doesn't support UNIX sockets. 

Finally the patch is wrong because its doing

+            if os.path.exists('/usr/bin/netcat'):
+                netcat = "netcat"

which is in the context of the machine virt-manager is running on, which
is totally bogus when connecting to a remote machine.
(Continue reading)

Permalink | Reply | 1 comment |
headers
Cole Robinson | 1 Dec 16:28
Picon
Favicon

Re: [PATCH] virtinst: also look at xorg.conf for keymap

Guido Günther wrote:
> Hi,
> attached patch looks at xorg.conf if no /etc/sysconfig/keyboard for the
> default keyboard. This helps with distros that don't have
> /etc/sysconfig/ like Debian.
> Cheers,
>  -- Guido
> 

Couple small nits:

>  HG changeset patch
> # User Guido Günther <agx@...>
> # Date 1227979760 -3600
> # Node ID a292eb8430cd2d9544e89dba127323704f3963ce
> # Parent  16e923676e84d81cf9bf6dc2cdd30d95bf65bbe1
> parse xorg.conf for keymap
> 
> diff -r 16e923676e84 -r a292eb8430cd virtinst/util.py
> --- a/virtinst/util.py	Tue Nov 25 10:25:08 2008 -0500
> +++ b/virtinst/util.py	Sat Nov 29 18:29:20 2008 +0100
> @@ -33,6 +33,7 @@
>  
>  
>  KEYBOARD_DIR = "/etc/sysconfig/keyboard"
> +XORG_CONF = "/etc/X11/xorg.conf"
>  
>  def default_route():
>      route_file = "/proc/net/route"
> @@ -276,6 +277,28 @@
(Continue reading)

Permalink | Reply | 3 comments |
headers
Cole Robinson | 1 Dec 16:33
Picon
Favicon

Re: [PATCH] virtinst: pool separator is '/' not ':'

Guido Günther wrote:
> Hi,
> Doc typo. Possible patch attached.
> Cheers,
>  -- Guido
> 
> 

Applied:

http://hg.et.redhat.com/virt/applications/virtinst--devel?cs=0a953a339693

Thanks,
Cole
Permalink | Reply | 0 comments |
headers
Cole Robinson | 1 Dec 16:44
Picon
Favicon

Re: [PATCH] virtinst: also look at xorg.conf for keymap

Cole Robinson wrote:
> Guido Günther wrote:
>> Hi,
>> attached patch looks at xorg.conf if no /etc/sysconfig/keyboard for the
>> default keyboard. This helps with distros that don't have
>> /etc/sysconfig/ like Debian.
>> Cheers,
>>  -- Guido
>>
> 
> Couple small nits:
> 
>>  HG changeset patch
>> # User Guido Günther <agx@...>
>> # Date 1227979760 -3600
>> # Node ID a292eb8430cd2d9544e89dba127323704f3963ce
>> # Parent  16e923676e84d81cf9bf6dc2cdd30d95bf65bbe1
>> parse xorg.conf for keymap
>>
>> diff -r 16e923676e84 -r a292eb8430cd virtinst/util.py
>> --- a/virtinst/util.py	Tue Nov 25 10:25:08 2008 -0500
>> +++ b/virtinst/util.py	Sat Nov 29 18:29:20 2008 +0100
>> @@ -33,6 +33,7 @@
>>  
>>  
>>  KEYBOARD_DIR = "/etc/sysconfig/keyboard"
>> +XORG_CONF = "/etc/X11/xorg.conf"
>>  
>>  def default_route():
>>      route_file = "/proc/net/route"
(Continue reading)

Permalink | Reply | 0 comments |
headers
Cole Robinson | 1 Dec 16:53
Picon
Favicon

Re: FYI: Changes to virtinst test suite

Guido Günther wrote:
> HI Cole,
> On Mon, Nov 24, 2008 at 05:24:45PM -0500, Cole Robinson wrote:
>> Hi all,
>>
>> I've added some changes to the virtinst test suite that will affect
>> anyone writing patches. Two commands have been added to the setup.py script:
>>
>> python setup.py check    : runs a pylint wrapper against the
>>
>>                            codebase
>> python setup.py test_urls: runs a url fetching test suite
> Cool thing! Would it make sense to release a new version? Hg acquired
> quiet some new features and bugfixes, so I'd probably make sense. Having
> virtinst release more often would help distros to get the new stuff out
> quicker.
>  -- Guido
> 

Completely agreed on releasing more often: the huge lapse between the
previous two releases was no good. I think pushing new virtinst +
virt-manager in a couple weeks would be fine, there are still a couple
other things I'd like to pick off until then. Going forward, my goal is
to have an actual schedule, and transparency wrt planned features,
timelines, etc.

Thanks,
Cole
Permalink | Reply | 0 comments |

Gmane