[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: remote install on 100+ workstations?



On Wed, Mar 14, 2001 at 02:46:48PM -0800, Chris Majewski wrote:
> The  computer  science department  at  my  university  has many  Linux
> boxes.  Say,  on  the  order  of  100.  Almost  all  these  boxes  run
> RedHat (not Debian, but read on). 
> 
> I don't like RedHat  that much: for example, RedHat 7.0 ships
> with broken  kernel headers, an unreleased and  unsupported version of
> the gcc compiler,  and a glibc version on which  gcc will only compile
> after applying patches (these  patches complicate life by changing, in
> an  architecture-dependent   way,  header  files  which   get  put  in
> architecture-independent places).  

And of course RPM's quickly become a pain in the ass.

> I would like to investigate replacing RedHat with Debian. 
> 
> The  current  rationale  for  using  RedHat is  that  there  exists  a
> mechanism for  installing/upgrading many RedHat boxes,  in a customized
> way, over the  network. This mechanism is called  "kickstart". I don't
> know much  about it. I  don't know if  a similar mechanism  exists for
> Debian. However, I suspect that it does. 

See http://www.informatik.uni-koeln.de/fai/

Dunno how well it works, since I haven't used it.

> My question, then, is this: does  anyone have (or know of) a mechanism
> which will  allow us  to install Debian  remotely on a  hundred+ boxes,
> including  department-specific  customizations,  such as  patches  and
> non-Debian  files?    Given  that  we  are   a  research  environment,
> administered by  some pretty  clueful people, our  linux installations
> will necessarily  be very customized. So something  like 'apt-get', by
> itself, is not good enough as I know it. 

Actually, once past the install phase, apt-get and things like debhelper
should be sufficient. :)

Set up your own repository containing the things that you want.  Instead
of (or in addition to) the 'main/contrib/non-free', make directories for
'webservers', 'develboxes', 'testing' or whatever, where you put in
your toys.  The machines pointing to 'develboxes', for example, would
get a copy of perl with all the debugging stuff turned on, while the
webservers would get a normal one, perhaps with some superduper
optimizing turned on.  The 'testing' is for the bleeding edge folks
who can expect things to break, or you could do that by creating a
seperate 'distribution' for them much like Debian proper does.  How
you break things up depends on your needs.

One of the beauties of Debian is that it is very easy to adapt into your
own distribution .... and if done right you can do it without breaking
compatibility with regular Debian.

-- 
CueCat decoder .signature by Larry Wall:
#!/usr/bin/perl -n
printf "Serial: %s Type: %s Code: %s\n", map { tr/a-zA-Z0-9+-/ -_/; $_ = unpack
'u', chr(32 + length()*3/4) . $_; s/\0+$//; $_ ^= "C" x length; } /\.([^.]+)/g; 



Reply to: