[linux-elitists] What to do about cluebatting such companies, that require possibly *YEARS* old Distros

Rick Bradley roundeye@roundeye.net
Wed Jan 26 13:14:08 PST 2005


* glen martin (glenm@locutory.org) [050126 15:54]:
> In fact, there are only two good times for large upgrades:
>   - when replacing the system, and this is rare. Many folks
>     haven't 'upgraded' from COBOL yet.
>   - when extending the system, in which case the
>     extension is coded in something new and it calls
>     the untouched old system.
> 
> Sure, folks will upgrade for security patches. Well,
> they'll select which ones they need, and which systems
> they're worried about. But change is a problem when
> lives and revenues are in the stream.

In a properly managed production environment software upgrades of any
size should not be an ordeal in the least.  One can run their own
package repositories for distro of choice, and segregate testing from
production for the various components being run.  Of course one's own
tagged software can be packaged and distributed as well.  Anything that
doesn't work on the testing environment shouldn't be pushed onto a
production system.  Make it simple to switch a "production pointer" from
one system (or cluster or network) to another, and back, and even
unexpected incompatibilities can be quickly reversed.  Testing becomes
production, and the old production system(s) become the new testing
system(s).

Then again, I suppose we're talking about improperly administered
environments almost by definition in this thread, so, um, carry on.

Rick
-- 
 http://www.rickbradley.com    MUPRN: 882
                       |  who like to have the
   random email haiku  |  old revision number in
                       |  the log message.



More information about the linux-elitists mailing list