You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@vcl.apache.org by Brian Bouterse <bm...@ncsu.edu> on 2009/03/17 23:12:40 UTC

Re: More questions about esx.pm

I've copied the vcl-dev list on my response because it's the  
appropriate place for VCL discussion.

The copying of the .vmx file to the image's golden directory when  
capturing files serves no purpose.  It will be removed eventually.

Creating .vmx files on the fly works well for us because it lets us  
size the RAM dynamically for each VM based on the database's image  
meta-data.  Similarly, we can connect the networking differently for  
different VMs based on meta-data (for example the virtual switches the  
VM should uplink to)

Best,
Brian


Brian Bouterse
Secure Open Systems Initiative
919.698.8796




On Mar 17, 2009, at 4:31 PM, Peng Ning wrote:

> Hi Brian,
>
> We got most of our questions resolved. In your current esx.pm file,  
> you generate the .vmx file on the fly when loading images, but copy  
> the .vmx file into the golden image when capturing images. What's  
> the benefit of creating the .vmx file on the fly? Does it fit the  
> working environment better? Are there reasons behind it?
>
> Thanks,
>
> Peng
>
> ================================================
> Peng Ning, PhD
> Associate Professor of Computer Science
> Technical Director, Secure Open Systems Initiative
> NC State University, Raleigh, NC 27695-8206
> Voice: (919)513-4457, Fax: (919)515-7896
> Email: pning@ncsu.edu
> http://www.csc.ncsu.edu/faculty/ning/
> ================================================
>
>
>
> On Mar 14, 2009, at 8:04 PM, Peng Ning wrote:
>
>> Hi Brian,
>>
>> I got more questions when I read the code. Hope you are not bored  
>> with my questions.
>>
>> In the load sub, you always create the vmx file by yourself. But in  
>> the capture sub, you copy (and fix) the vmx in use to the golden  
>> image directory. This seems unnecessary at first glance. Is there  
>> any specific reason?
>>
>> You copy the vmdk and vmx files to /mnt/vcl/inuse/ 
>> $computer_shortname (i.e., $datastorepath4vmx), but ssh-copy the  
>> flat.vmdk file to /mnt/export/inuse/... (i.e.,  
>> $datastore_share_path). Will do they seem to be different  
>> locations? How do the target blade access these locations? I guess  
>> $datastore will be remotely mounted on the target blade. Is  
>> $datastorepath4vmx also mounted remotely on the blade?
>>
>> Thanks,
>>
>> Peng
>>
>> ================================================
>> Peng Ning, PhD
>> Associate Professor of Computer Science
>> Technical Director, Secure Open Systems Initiative
>> NC State University, Raleigh, NC 27695-8206
>> Voice: (919)513-4457, Fax: (919)515-7896
>> Email: pning@ncsu.edu
>> http://www.csc.ncsu.edu/faculty/ning/
>> ================================================
>>
>>
>>
>