Automatic update

[John Summerfied]

Paul Michael Reilly wrote:
> One of the original reasons that I chose Redhat/Fedora was the ability
> to learn easily that an update was available and/or necessary and get
> that update installed.  I liked the panel applet notification but that
> feature has been a rocky road, it used to work, didn't work and who
> knows at any given time what it does now.  So I've pretty much given
> up on the panel notification applet.  Red Hat update was equally rocky
> with repo hell mimicking dll hell on Windows.  That has gotten better
> over time (with Yum, haven't used Red Hat update in ages).  Now my
> wife points out that Microsoft has a painless and practical update
> facility that appears to be what I was looking for from Linux.  So my
> question is: why has the Redhat/Fedora update facility fallen into
> such disrepair?  And is there any reason to expect that it will become
> competitive with Microsoft's with FC5 and follow ons?.  It would be
> unfortunate to hear that the answer is Redhat Enterprise versions.  I
> don't mind paying but I do mind running stale software.  I'll pay
> gladly to live on the bleeding edge.  Even the lack of Firefox 1.5 with
> FC4 is disturbing, enough to make me wonder what all the fuss over
> Ubuntu is all about.

Franky, I don't trust anyone's automatic update.

I am perfectly happy to download the updates automatically, where this 
is feasible (ie, not dialup and not mobile), but _I_ choose when to 
apply those updates.

Microsoft's Windows Update has broken systems in the past, and is highly 
likely to do so in the future.

The Fedora project has released kernels for Fedora Core 3 to my certain 
knowledge that lead to broken systems, and I'm almost certain it did so 
for FC4. I know kernel updates are disabled by default - it's just an 
example, there are other packages such as glibc that also can completely 
break a system.

It's entirely natural that people make mistakes, many (hopefully minor) 
mistakes in any given day. Often, the difference between a small mistake 
and a great one is when it's detected.

The sensible defence against such mistakes includes countermeasures such 
as _not_ using automatic updates, waiting a while with the angels for 
the fools to rush in, and to review the updates before applying them so 
as to guage the possible damage.

Those are the minimum; larger enterprises and govenment departments are 
like to have test machines where they install these updates and test 
their regular workloads against them, and only then if they check out 
okay do they roll them out to the users. This applies whether it's 
Windows or Linux clients or servers, IBM's iSeries and pSeries Power 
machines running OS/400 (or whatever its called now) or AIX, or IBM 
z/VM, Z/OS or the latest incarnation of VSE or even Linux: mistakes are 
likely to cost millions, and they take it very seriously indeed.


-- 

Cheers
John

-- spambait
1aaaaaaa at computerdatasafe.com.au  Z1aaaaaaa at computerdatasafe.com.au
Tourist pics http://portgeographe.environmentaldisasters.cds.merseine.nu/

do not reply off-list