[NCLUG] What's happening with CentOS?

[R P Herrold]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Mon, 4 Apr 2011, dann frazier wrote:

> On Mon, Apr 04, 2011 at 02:50:56PM -0600, Bill Thorson wrote:

>> I'm curious what is happening in CentOS world.  I was perusing the
>> centos.org site today and noticed that almost nothing has changed
>> there in months or even years.  They have also not come out with a

> LWN had a article about this recently:
>  http://lwn.net/SubscriberLink/435744/2d6273bee714fbb1/

The article is long on speculation, based on reading selected 
threads off the CentOS mailing lists.  It is not however 
reporting, unless reporters simply do not talk with primary 
sources, these days

I call this the: 'latest is greatest' disease.  Just because 
someone suggests something does not mean it is a needed 
feature-set.  My question at the threshold would be: what of 
the upstream's version 6 is a needed feature [there _is_ mneat 
stuff in there, but nothing really earth-shattering] ... I'll 
address this later in this piece

I've done local testing builds of what the sources that will 
become CentOS 6have been done for over three months.  The 
sequencing issue really stems from the fact that Red Hat 
dropped 4.9, 5.6, and 6 initial in close succession, and the 
CentOS team made choices on what to 
ship first, falling to the side of: support our fielded boxes 
first (CentOS) vs. SL's release of its 6 sources rebuild

SL made a bit of a public misstep of not including needed 
files: .discinfo and .treeinfo files on their ISOs that 
prevented media based installs (coming from a lack of folks 
and process to do the QA coverage CentOS does) I think the 
world of Troy and Connie's work at SL [a team of two, not four 
as one comment in the LWN article suggest] -- full confession, 
and SL is not alone: we at CentOS were at the doorstep of a 
public release with a similar defect early in the 5 series 
that was caught 'at the last minute' and some mirrors 
'cheated' and released [locally flipped on the +r bit] a 
version that the CentOS project itself withdrew.  Shame on 
those mirrors that jumped the gun

The 5.6 updates and new install ISO images should land this 
week [they provisional final bits are percolating out to the 
several hundred mirrors as I type this] CentOS 6 had, as I 
recall, about 60 unsolved packages, and the anaconda changes. 
I expect CentOS 6 to issue Sometime in April, barring some 
issue in some corner case leaf node package, seems reasonable 
and likely

The Anaconda developers think that GUI installs are the way to 
go, and that text mode installs are deprecated.  I hit some 
problems needing to fall back to the VESA mode video drivers 
rather than the mis-detected one selected.  WHile I worked 
around this, I am not sure that the CentOS team won't get 
'stuck' on something that the upstream did not see fit to 
solve before shipping [I've had an open bug on this 
mis-identification of a common video Intel chipset in Red 
Hat's bugzilla for over 8 months ;( ]

I've commented on the CentOS mail list to the effect that I am 
running (and have been running) a trial build of the Red Hat 
sources as needed for my purposes.  Feels a lot like 5, and 
like 4, and like 3, and like 2.1, and like the RHL series 
before it.  Hardware compatibility profile between 5.6 and 6 
initial should be basically identical (5.6 is actually 'cut' a 
bit later and so a bit more).  But building a long lived 
enterprise service unit is not going to be bleeding edge 
exotic hardware, anyway

I was surprised to find a Broadcom wireless network card 
driver I need entirely removed from the upstream's 6, and 
pleased to find the 'atl1c' network driver I need for one unit 
present.  As I sort out why it seems to be a license issue as 
to the first ... ;(

- -- Russ herrold
 	herrold at centos.org

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (GNU/Linux)

iD8DBQFNmosWMRh1QZtklkQRAivFAJ4wISGn9GTHzPUnIAxPZyk1mVCzhACePMAF
aNRToZ2RqRA2L2sOENN5UbE=
=IDhs
-----END PGP SIGNATURE-----