Subject: Mandrake 9.1 Install (review? definately a long one..)
From: Leif Sawyer (lsawyer@gci.com)
Date: Thu Mar 27 2003 - 09:24:15 AKST
Well, I installed Bamboo on a couple of machines, and thought I'd go
over some of my issues:
Machine #1
IBM ThinkPad 760ED with p5-133
80 Mb Ram
20 Gig HD
mwave modem/sound
trident vga
3c5x9 PCI 10/100 in docking station
4x IDE cdrom
PS/2 Mouse
Machine #2
Tyan Dual CPU m/b with 1 p6/Celeron 1G
512 Mb Ram
20 Gig IDE HD
2 x 80 Gig IDE HD
4/24/2 IDE CDR/RW/DVD
nvidia g-force 2MX
SoundBlaster Live
3c905 PCI 10/100
USB IntelliMouse+Wheel
I installed on Machine #1 first, as I had the least amount of personal
data on it -- simply copied it over to my Win95 partition and started
the
install. I have to use a UniversalBoot floppy in order to boot off of
the
CDROM (which i find easier than trying to make a boot floppy) and this
worked well. Selected expert mode.
First thing I noticed is that the graphical installer seems to use
32-bit color,
and as my card doesn't support that many colors, it looked a little
funny. I didn't
bother rebooting into vga-low mode, but just suffered through it. Not
unreadable,
just not nice-looking.
The new installer is very slick. Much cleaner than previous versions. I
quickly
re-formatted my linux partitions and was on the way to installing the
OS.
The installation completed without a hitch, and I rebooted my laptop.
I copied my original XFree86cfg file over, changed my default window
manager
from KDE to IceWM (save on some resources) and fired up X.
Total time: about 3 hours (mostly due to slow machine!)
Next I upgraded machine #2. This machine is more complicated, due to
the
/usr partition being a raid-1 meta-device of two partitions (hdb1 +
hdc1)
I booted the CD into rescue mode, mounted my root drive and copied over
the
raidstart/raidstop binaries, as they weren't on the rescue image (bad
mandrake!)
I then had to manually create the /dev/md0 node (mknod /dev/md0 b 9 0)
I also copied my raidtab into the ramdisk /etc. I loaded the raid1
drivers
and started raid. mounted /dev/md0 on /mnt/usr, and proceded to rm -rf
everything that I didn't need. (this machine had been upgraded since
mdk 7.1, it
was past due for a wipe.. :-)
once finished, i mv'd the /mnt/etc to /mnt/etc.save, umounted /mnt/usr
and /mnt,
raidstop'd, and rebooted.
I chose the expert installer again, and when it came time to choose my
partitions,
i selected my existing root partition, then switched to the console.
I again had to cp my raidtab into /etc, create the /dev/md0 node,
modprobe raid1
and raidstart. raidstart/stop are on the install image, however.. I
then
mounted /dev/md0 on /mnt/usr
switching back to the graphical installer, I picked my options and let
run. 20
minutes later, the packages were installed, and it asked me for my
password.
After defining the root and user passwords, it then proceded to the
network setup,
which is where I had my first hang -- insmod 3c59x was stuck
initializing.
I switched to the console, umounted /mnt/usr and /mnt, raidstop'd, and
shutdown.
I removed the ethernet card and restarted the install (same steps as
previous)
choosing to upgrade my station (as it detected a mandrake install) and
it skipped
ahead and seemed to complete the install without any further user
interaction
(bad mandrake, as you'll see)
At the 'reboot' screen, i switched back to the console, umount'd and
raidstop'd
before letting mandrake reboot.
you may be familiar with this: "LI 99 99 99 99 99 99 99 99 99 99 99 99
[....]"
Sigh.
Boot back into the rescue disk, mount my root, discover that some
symlinks are
missing/incorrect, there is no lilo, no initrd. Mount my /usr
partition,
mkinitrd, install lilo, shut everything down and reboot.
This time, I was greeted with a login prompt for "localhost". Ok, no
networking
config. I shutdown, install my network card, and start back up. log
back in
and discover that there is no /usr mounted. edit /etc/fstab, copy
/etc.save/raidtab
to /etc, and reboot.
(are you counting? almost 10 reboots so far.. this is like windows!)
login, fix /etc/modules.conf so that my network card is correct, copy my
saved
/etc/sysconfig/network, copy my saved
/etc/sysconfig/network-scripts/ifcfg-eth0,
and run ifup eth0. Phew. network connectivity.
start x -- oops! no x server. mount the disk, install XFree86-4
server. copy my
saved XFree86 configs. try again. Oop! no nvidia module! make the
nvidia modules
and start x again. bingo!
configure my printer via the cups web interface. bingo!
try xmms.. bzzt! not working. My sound looks configured right, but I
still
haven't been able to get it working. (bad mandrake!)
configure httpd2 to use my old webroot and server certs. bingo!
(realize that I have to re-install the PEAR db modules for php and
restart.. heh)
try to start postfix. bzzt! can't find /usr/libexec/postfix... seems
it's
been moved to /usr/lib/postfix, but they didn't tell the binary that.
Make a symlink
and restart. (bad mandrake!) bingo!
Elapsed time to a (mostly) fully functional system?
30 mins + 3 hours of diagnostic cleanup and workarounds.
Conclusions? Expect a mandrake 9.2 shortly which will clean up some of
these
issues. But so far, it's been running pretty well. The 2.4.21(pre)
kernel
seems much more responsive than the 2.4.19 previously used. KDE 3.1 is
nicer as well. I haven't really played much with it, but it's a little
snappier.
As soon as I get sound working, i'll start trying games again. I'll
have to
recompile vmware for this kernel, and see how well it runs too.
Leif
( score: 8/10, 4/10 == 6/10 )
-- Binary/unsupported file stripped by Listar --
-- Type: application/x-pkcs7-signature
-- File: smime.p7s
---------
To unsubscribe, send email to <aklug-request@aklug.org>
with 'unsubscribe' in the message body.
This archive was generated by hypermail 2a23 : Thu Mar 27 2003 - 09:24:20 AKST