SUMMARY: Re: cloning solarins installations

On Wed, Nov 30, 2005 at 08:34:04PM +0100, Karoly VEGH wrote:
> On Mon, Nov 21, 2005 at 10:13:57PM +0100, Karoly VEGH wrote:
>  
> > I would like to clone solaris installations - e.g. I would like to
> > remove one disk from a complete installation from a raid1 device, 
> > put it into another machine, boot from it, sync it up in raid1 again
> > with a third disk.
> 
> Hi, 
> 
> I am about to solve this problem, though to be able to summarize I
> sould once more need your kind help. 
> 
> The problem is: I do not want to halt the server, to remove one disk, so
> I created a flash archive with "flarcreate", now I would like to setup a
> tftp and bootserver, to be able to boot the other boxes over the net,
> and during the installation use the flash archive shared on NFS. 


Dear sunmanagers members - thanks for all the help. The problem was
again a PEBKAC-type (problem existed between keyboard and chair). 

Of course a problem like this is hard to detect remotely, but still I
found it, and your support was a big help in this. Thanks. 

The problem was, that I had a bad routing table, routing internal
addresses defaultly through the external interface... kind of
embarrassing :I - as soon as I fixed that, everything worked right
witha jumpstart installation.

Still, I got really good suggestions, ones that I haven't been aware
of, let me share: 


The solutionsuggestions were: 


Darren Dunham, Dan Stromberg and Hutin Bertrand suggested the same: 

> Does in.tftpd run from inetd at that time on the server? 
> What happens when you try to retreive it directly on the server?  
> # cd /tmp   
> # tftp localhost   
> tftp> get C0A80002 

------

Clift, Robert suggested controlling the consistence of /etc/ethers /etc/hosts 
and /etc/bootparams.

------

Saxon, Stuart informed asked/suggested: 

> Where did you get the mac address from {banner at OBP}                      
> 
> and is               
> 
> within the OBP is local-mac-address set to true - it should be set to false

Thanks - that was wrong too. 

----- 

joe fletcher, Vladimir and Paul Greidanus suggested using jumpstart/flar utility (which 
was the endsolution): 

> Yesterday I've successfully cloned two Solaris 9 nodes with "flar" utility, 
> which appeared in Solaris 8. You can find mode about it:                    
>                                                                             
> http://www.sun.com/bigadmin/content/submitted/flash_archive.html            
>                                                                             
> The only unhandy thing - you need to recreate raid-1 on cloned node once    
> again.

-

> It's possible, but I can't help on that end. Jumpstart shouldn't take       
> more then a hour or two to get working, and it's a little more flexible     
> if you need to clone more then one machine.                                 
>                                                                             
> Actually, the solaris installer CD should allow you to choose a flash       
> archive installation, so you don't necessarily need a jumpstart server,     
> just a flash archive somewhere on NFS.


They were right - setting up jumpstart doesn't take that long, and
though it wasn't the way I wanted to solve the problem, it was surely
the easier path. thanks.

-----

> Santhakumar, Siva suggested, that if I erally want to do this, I have to detach the 
> devices, and clear a metadb

I didnt come to that, btu next time I will definitely try it.


with best regards, 


charlie

-- 
Vigh Karoly  -  EUnet Telekom GmbH  -  Team Systems
Nussdorfer Ldnde 23, A-1190 Wien, Vienna, Austria
http://www.eunet.at Tel: +43 (0) 591590 / Fax: +43 (0) 591593001
see Disclaimer http://www.eunet.at/signatur/
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers