Hi All,
Our Windows team predominantly use Ghost to deploy default desktop images. They have been toying with migrating across and using PING (http://ping.windowsdream.com/) instead for a variety of reasons, mostly to do with the general nastiness of getting newer hardware to play with the DOS based Ghost environment.
After having a quick look at PING (not used it myself as of yet), I was wondering if I could re-purpose parts of the Cobbler infrastructure (PXE management, etc), so that it can use PING as a means of creating a Windows provisioning framework for our default images.
Obviously this isn't quite the same as kickstart and cobbler, (due to the need to image a golden-image system first), but I presume it'll work OK (with the caveats of using images). After reviewing the System Retirement with PXE and DBAN article on the wiki, it should be doable???
Does this all sound sensible??? I think so, but as I haven't done much Windows deployments for a few years, its all a bit hazy in my mind.
Last but not least, has anyone done this before and if so any tips/tricks to getting it going or was it all straight forward enough??? Otherwise I'll make a stab at it, see what happens and document any success.
Thanks
Dan
-- Dan Hawker d.hawker@kew.org UNIX & Comms Royal Botanic Gardens - Kew --
"I love deadlines. I like the whooshing sound they make as they fly by" - Douglas Adams
--
Daniel Hawker wrote:
Hi All,
Our Windows team predominantly use Ghost to deploy default desktop images. They have been toying with migrating across and using PING (http://ping.windowsdream.com/) instead for a variety of reasons, mostly to do with the general nastiness of getting newer hardware to play with the DOS based Ghost environment.
After having a quick look at PING (not used it myself as of yet), I was wondering if I could re-purpose parts of the Cobbler infrastructure (PXE management, etc), so that it can use PING as a means of creating a Windows provisioning framework for our default images.
Provided it is available under an open source license, yes. It seems to be based on LFS, so I would hope yes, though their web site is not very explicit.
The new "cobbler image" objects in 1.1 seem to be a good candidate for building something like this around.
As long as the workflow for generating an image is fairly straight forward I don't see any problems with this offhand. It's not really Windows specific is it? Partimage isn't.
Obviously this isn't quite the same as kickstart and cobbler, (due to the need to image a golden-image system first), but I presume it'll work OK (with the caveats of using images). After reviewing the System Retirement with PXE and DBAN article on the wiki, it should be doable???
Does this all sound sensible??? I think so, but as I haven't done much Windows deployments for a few years, its all a bit hazy in my mind.
It does to me. Typically with Windows boxes I believe people image them and then do something to replace the UUID.
Last but not least, has anyone done this before and if so any tips/tricks to getting it going or was it all straight forward enough??? Otherwise I'll make a stab at it, see what happens and document any success.
I think the first steps is to get a set of instructions together about how to clone Linux and Windows systems with this, and how they would integrate with a generic PXE environment. You can achieve this by doing simple things like editing /etc/cobbler/pxedefault.template to add entries into the PXE menu, and to copy images into /tftpboot/myimages (or /var/lib/tftpboot/myimages for F9) and seeing how that works.
If we can get a general workflow going with that, we can see what it would take to make it even more automatable via "cobbler image" type commands -- and whether this makes sense to add something around or if just documenting details is enough. I think being able to map a cobbler system to a image that represents a bare-metal install of a foreign OS /is/ interesting, so I think the answer is yes.
This is definitely something I am interested in, and if it makes sense (and ping isn't really windows specific), we can look at getting it packaged in Fedora as well. Mainly we do want to make sure
Folks who would like to propose other alternatives are welcome to do so. I've asked about this a bit before on list and we didn't really come up with any perfect solutions, so there is probably some more room for playing around with ideas.
Thanks
Dan
-- Dan Hawker d.hawker@kew.org UNIX & Comms Royal Botanic Gardens - Kew --
"I love deadlines. I like the whooshing sound they make as they fly by" - Douglas Adams
-- _______________________________________________ cobbler mailing list cobbler@lists.fedorahosted.org https://fedorahosted.org/mailman/listinfo/cobbler
Perry N. Myers wrote:
Michael DeHaan wrote:
It does to me. Typically with Windows boxes I believe people image them and then do something to replace the UUID.
The tool to do this is called Sysprep and it replaces the system id (SID) and generates a random computer name.
Perry _______________________________________________ cobbler mailing list cobbler@lists.fedorahosted.org https://fedorahosted.org/mailman/listinfo/cobbler
Appendum:
There's also some very promising method I've heard about using liveish images, PXE deployed, to clone via systemimager to an NFS share.
There's someone looking at that, so if we do this, we have Windows on metal essentially solved using entirely free OSS tools in a way that is not Windows specific. Joy!
--Michael
Michael DeHaan wrote:
Perry N. Myers wrote:
Michael DeHaan wrote:
It does to me. Typically with Windows boxes I believe people image them and then do something to replace the UUID.
The tool to do this is called Sysprep and it replaces the system id (SID) and generates a random computer name.
Perry _______________________________________________ cobbler mailing list cobbler@lists.fedorahosted.org https://fedorahosted.org/mailman/listinfo/cobbler
Appendum:
There's also some very promising method I've heard about using liveish images, PXE deployed, to clone via systemimager to an NFS share.
There's someone looking at that, so if we do this, we have Windows on metal essentially solved using entirely free OSS tools in a way that is not Windows specific. Joy!
--Michael
s/systemimager/partimage/
cobbler@lists.fedorahosted.org