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

[Sheflug] Re: Updating a Mixed Network - or Just Linux ?



>>>>> "Richard" == Richard  <richard [at] sheflug.co.uk> writes:

Forgive me if I got Owen's words and Richard's mixed....

    Owen%R> I'd like a utility which I can configure to manage all of
    Owen%R> these machines at once, in that I can say to it `On any
    Owen%R> machine which is running exim 3.13, please replace it
    Owen%R> with exim 3.16' and be certain that exim will not get
    Owen%R> installed on machines that do not have exim running.

I know a number of people doing this kind of thing, but don't want to
be identified (silly corporate policies).  You're welcome to come to
Tokyo on Dec 15 and pour fizzy truth serum down their throats at the
TLUG end-of-year party, though ;-)

Anyway, what they have said (this was about a year ago) is that the
only way they've found to do this is make a class of machines that all
are configured identically, and put /usr on an NFS or AFS mount.

The "Depot" system works this way IIRC.  Depot was originally intended
for heterogeneous platforms (eg, installing Perl simultaneously on i86
and Sparc platforms, with the binaries automounted from arch-specific
hierarchies, and the Perl code shared).  The GNU /usr/local manager is
more like Webmin, I think.

If you don't like the network FS aspects, Coda should work too, with
really high hoard priorities on system files, subject to the caveat
that nobody has really tried /usr on Coda yet AFAIK.  (It would have
to be some kind of symlink hack since Coda probably won't be happy
mounted anywhere but /coda.)  Alternatively, you could have have the
model machine export its /usr filesystem, mount it on /model/usr, and
just cp the whole thing across on updates.

    Owen%R> Also I want it to make `sensible decisions' (however that
    Owen%R> is defined!) about things like changes to config files.

On Debian, there is no such thing as a "sensible decision" algorithm;
you have to look at every change.  Sometimes accepting the changes
breaks things, sometimes refusing them does :-(

This "sensible config" probably would break all of the methods I
suggested anyway, if there are host-specific configuration.

-- 
University of Tsukuba                Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
Institute of Policy and Planning Sciences       Tel/fax: +81 (298) 53-5091
_________________  _________________  _________________  _________________
What are those straight lines for?  "XEmacs rules."
---------------------------------------------------------------------
Sheffield Linux User's Group - http://www.sheflug.co.uk
To unsubscribe from this list send mail to
- <sheflug-request [at] vuw.ac.nz> - with the word 
 "unsubscribe" in the body of the message. 

  GNU the choice of a complete generation.