Kock, Boris - Acision wrote:
> Hi all,
>
> Recently I discovered cobbler. First of all, a word of thanks for creating 
> this great product.
>
> Next to cobbler I have also used koan to deploy vmware server guest systems 
> on vmware server. I tried it with vmware server 2.0 to see if it would be 
> supported ... as you might know it is not (yet), since the command line 
> interface to create virtual machines has changed. With some 'hacking' it is 
> possible to get it to work. However I broke the vmware server 1.0.x support 
> with that.
>   

Is it possible to test for the version inside of the koan/vmwcreate.py?  
I'd think it would be.
> Before starting the task of integrating this properly in the koan code a 
> couple of questions, based on what I noticed with testing:
>
> 1. On a number of places in the koan code checks are done for the libvirt 
> api. Reading the libvirt documentation I did not find any reference to 
> support for vmware now or in the future. 


Correct... not all install types in koan use libvirt.  We wish they all 
could, but vmware is a special case (and well, it's also proprietary, 
but widely used enough that I think making Linux easier to install on it 
is goodness)
> In order to be able to use koan on systems < FC8 to install vmware guest it 
> makes sense to remove these checks for vmware actions. Is there a reason for 
> the checks in the current code?
>   

Ah, I think I follow, you're saying that libvirt must be installed even 
if you're using vmware.   Yes, we should bypass those checks if the 
virt-type is not libvirt based.   100% correct.
> 2. Since the command syntax changed for vmware server it is necessary to be 
> able to make the distinction between vmware server 1.0.x and vmware server 
> 2.x. The best solution in my opinion seems to spli the --virt-type=vmware 
> option in --virt-type=vmwares1 and virt-type=vmwares2. Does this make sense?
>   

If we can't do it at runtime on the system, that's ok by me, though we 
should probably make a "vmwares2" and leave the original as the "1" for 
backwards compat reasons.  Ideally if we can query the package to find 
out (even the RPM version) that's probably a better way to go.

> 3. Is there a reason why the command syntax and help are different for 
> cobbler and koan?
>   

The help is different in cobbler because we have multiple layers of 
subcommands before we get to optparse, but when we get to the options, 
both use optparse.

Is there another difference you might be pointing out?

Additionally, I should point out that we do also have a VMware 
workstation virt type -- we've still never done ESX.   If you're 
interested in that, that would be an awesome module to add.  If not, 
this is something we'll probably look into for a future release.

Thanks!

--Michael
_______________________________________________
cobbler mailing list
[email protected]
https://fedorahosted.org/mailman/listinfo/cobbler

Reply via email to