Re: [PATCH] xenctld - a control channel multiplexing daemon

[Daniel Stekloff <dsteklof@us.ibm.com>]

On Wed, 2005-01-26 at 14:57, Anthony Liguori wrote:
> On Wed, 2005-01-26 at 15:49, Michael Hohnbaum wrote:
> 
> > While beyond the current focus, persistent store could feasibly
> > be used to hold domain definitions for non-existent domains and 
> > suspended domains.  One could envision adding a state field into
> > the domain configuration/definition.  Valid states for current 
> > capabilities would be {active, suspended, migrated, inactive}.  On
> 
> Yes, but the problem that occurs is uniquely identifying a domain.  In
> other words, what's the key used within the persistent store?
> 
> If it's domain id (which is what I assume it's going to be), you cannot
> tag it as having an "inactive" state because there's nothing that
> prevents a domain from being created with the same domain id.
> 
> Also, if you try to assign domains UUIDs or something, what do you do
> for cloning/checkpointing?  Do you assign a new UUID on a clone but not
> on a checkpoint?  Does assigning new UUIDs propagate to things like MAC
> addresses or other things that are supposed to be unique?


I think checkpoints should be stored by the domain id/key for the domain
they are from, using date as an identifier. If you're going to run that
checkpoint image, it then gets a new id to run under and a new entry as
a domain. 

Sorry about blasting out messages, just thinking about the questions
everyone has raised.

Thanks,

Dan 



> There's a lot to be thought about.  I think punting the problem (as Andy
> suggests) is the right approach for now.
> 
> Regards,



-------------------------------------------------------
This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting
Tool for open source databases. Create drag-&-drop reports. Save time
by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc.
Download a FREE copy at http://www.intelliview.com/go/osdn_nl