Having got my system fully working with RHL 9, only 2 months ago, I
decided that it was time for me to give something back to the community.
So I decided to join the Severn Beta trial.
Well here are my finidings:
As part of the install, I performed the CD check, even though CDRecorder
did not show any messages. It said the disks were fine, so I went ahead
with the install. However when the installer got to disk 2, it would not
read one of the packages, which meant I had to exit at this point,
reburn the CD and try again. Luckily it worked OK the second attempt.
Question : Why did the CD Check say there was nothing wrong with disk 2,
when clearly there was.
I decided intially it would make better sense to perform an upgrade
rather than a full install, since I had everything working.
This was my first mistake.
Once the installation was complete, I found that the system did not make
use of the new graphical boot. I assume the installer prefers to keep
the existing config in its entirity.
Once booted, I attempted to install the NVidia driver, rebuilding the
kernel. This failed as the NVidia driver performs a check of the CC
versions and found that they are compiled using an earlier release. I
assume NVidia will update there drivers when Severn goes into GA.
OK, so using the generic drivers, I logged into Gnome, attempted to make
use of Mozilla. Complete failure, no messages, just would not start.
I then attempted to resize the screen. The system at this point, did not
recognise the higher resultions, and only offered me a maximum of 800 x
I then noticed that even though I have only installed possible 4
additional RPM's ontop of the original RHL 9, I had alot of package
Anyway, this went on for a bit, so I decided to go for a full reinstall
of the RHL Severn Beta.
Once this was complete, the graphical boot appeared, which I have to say
slows the boot process down quite considerably, at least twice as long.
All of the applications started to work again.
Two configuration problems. My DVD+RW stopped working, due to the
/dev/cdrom being pointed at /dev/hdc when it should have been /dev/sdc0.
Even though IDE-SCSI was installed, it did not fully configure the
system, when the RHL 9 did. The other problem is related to my sound
card not being noticed, which it was with RHL 9. Not got down to the
bottom of this one yet, but will in time.
The screen resolution program picked up the correct list of resultions,
however when I increased the resolution, I noticed the background does
not get correctly painted, and the apps running still think that the
screen size is alot smaller, which looks perculiar when the applications
are maximised, and only using the top left corner. Also resizing the
window, passed the old maximum limit, causes the window to start
painting in a very strange manner, effectively picking sizes out at
random. All of which clears up after a restart of X.
I noticed on the forum a comment about the fonts. Even though they look
clear, they do look as if they have been streteched to a thin version
when compared to RHL 9.
Nice to see alot more of the latest versions of software.
Do you think that version 1.1 of OpenOffice will be included before the
Initial feelings are mixed, but the final outcome was good. Still got a
long way to go though.
refusing to defer to discretion (or good judgment), i'm going to take
one more stab at severn on my system, and i'd like to clarify some simple
issues regarding versions of gcc. one step at a time, i figure.
immediately upon installation, the kernel that is installed was clearly
built with gcc 3.2, while gcc itself is gcc 3.3 (although gcc 3.2 is still
available through the "gcc32" executable). this has a couple of immediate
first, there is the occasion where you have to deal with a kernel and
compiler mismatch, particularly with nvidia drivers. if you try to
compile the latest nvidia driver normally, you get a "mismatch" error that
you can override by defining the "IGNORE_CC_MISMATCH" environment
variable when you build. fair enough, but of course, you never know
when this issue will come up again elsewhere.
second issue is whether there is any problem simply recompiling the
installed kernel source with gcc (3.3, that is)? is there anything about
that kernel that's necessarily gcc 3.2-related?
i decided to test recompiling with gcc 3.3, using the config file found
in /boot, and the compile failed partway through the modules. rerunning
the make with "make CC=gcc32 modules", however, seems to be working. i
could, of course, just config out the offending (I2C) option, and try
again with gcc 3.3, which i'll probably try just for the entertainment
but is there any reason *not* to rebuild the kernel using gcc 3.3, in
order to avoid any future mismatch issues?
If one doesn't have the third CD, the install aborts in a quite
unpleasant manner. I was forced to power-down the computer (and then
re-install another system) because I had left the final CD at work. My
expectation was that this would result in some capabilities being missing.
This sounds like a great time saver... thanks for the tip.
From: Robert P. J. Day [mailto:firstname.lastname@example.org]
Sent: Tuesday, July 29, 2003 10:42 AM
Subject: Re: RHL Severn - Initial Findings
On Tue, 29 Jul 2003, Barry K. Nathan wrote:
> On Tue, Jul 29, 2003 at 05:26:33PM +0100, Matthew Winter wrote:
> > did not show any messages. It said the disks were fine, so I went
> > ahead with the install. However when the installer got to disk 2, it
> > would not read one of the packages, which meant I had to exit at
> > this point, reburn the CD and try again. Luckily it worked OK the
> > Question : Why did the CD Check say there was nothing wrong with
> > disk 2, when clearly there was.
here's what i've settled on as a technique for convenient installs.
first, *if* you can arrange it on the system you're using, create a
separate partition, perhaps at the end of the disk (out of the way), and
download your ISO images there. once they're there, and if you're
running on a current version of RH, you can mount them, poke around,
"md5sum" sum them to check their health, and so on.
once you're set up like that, you can either create a boot floppy or,
alternatively, burn a copy of CD 1 and, when you start the install, use
boot: linux askmethod
select "hard drive" install, point the installation program at the
/dev/hdxy partition and directory the ISOs are in, answer all the
standard questions, and walk away.
1) you can download, verify and examine the ISOs at your leisure
before starting the install, so you *know* they'll be healthy,
2) you don't need to burn any physical CDs (except, possibly, for the
first one just to start the install process)
3) you don't need to hang around and switch CDs -- the install program
will automatically move from one ISO image to the next
4) chances are, the install might be faster than normal given hard drive
speed versus CD speed (but no guarantees there)
and once you've installed, you have permament access to the CD
contents by creating a number of directories as mount points and
mounting the ISOs using the loopback feature. (i've actually added the
mount points for the ISOs into /etc/fstab so they're always available.)
anyway, just my $.02.
Rhl-beta-list mailing list
It really bothers me that there is no way to test X config in anaconda. If anaconda doesnt probe your settings correctly, and/or you don't set them correctly, it becomes a challenge to get past firstboot.
Anyone else have thoughts on this?
During the install of Severn, the kernel detects (right before anaconda gets
hda: hard drive
hdb: cd rom
hdc: zip 100
hdc: lost interrupt lost
hdc: lost interrupt lost
hdc: lost interrupt lost
Only until I put an zip cartridge in does the installation proceed. This
may be a problem for other removable media detections as well.
Network printing in Severn:
Both the Networked JetDirect and the Networked Novell (NCP) appear to be
broken in this version. There don't seem to be any details to include.
With Networked JetDirect the error message is:
There was a problem sending CUPS test page to 'lp357m' queue:
lpr: unable to print file: server-error-service-unavailable
(I am less certain about just which printer queue name to use with NCP,
so that may possibly be working, even though it appears to also be
broken in a similar way.)