Upgraded RH8.0 to ProWSv3... ahem, all did NOT go well
 

It all started yesterday, innocently enough, when I attempted to upgrade my Red Hat 8.0 system, running perfectly, to Red Hat Pro Workstation v3. I've been an RHN subscriber for years and maybe I didn't realize that I had it so good.

(After waiting on the phone for 20 minutes for someone, obviously in Bangladesh somewhere ... :mad: ... to inform me that no one could be bothered to answer my question, I've stumbled upon this blog. ( Whew! :cool: )

(1) At first, the ProWS disk wouldn't even recognize that I had a previous version of RH installed. I finally found linux upgradeany and it proceeded to install ... everything. Three CD-ROMs worth. It didn't even give me a chance to choose.

(2) Upon reboot, I find that I'm still running:
"Linux version 2.4.20-28.8 (bhcompile@daffy.perf.redhat.com) (gcc version 3.2 20020903 (Red Hat Linux 8.0 3.2-7)) #1 Thu Dec 18 12:53:39 EST 2003"

(2a) Looking at the /boot directory, I see a vmlinux-2.4.21-4.EL file, and the corresponding vmlinuz, but no initrd...img file. Furthermore, attempting to use /sbin/mkinitrd informs me that there is no initio module for the release; and so it is: there is not. Why not?

(2c) I also observe that no entry has been made for this kernel in grub.conf. So that's why I'm still running an 8.0 kernel. Good thing too I guess because if there's no initrd I don't think I'd get very far? But I thought that I'd just spent four hours installing an upgrade!

---

Think we be done yet? Oh, noooo.... :rolleyes:

---

(3) What lead me down this primrose path in the first place was the fact that up2date would not run. It crashed with an "error in pthread"... the exact details of which momentarily escape me. But basically, nothing worked. After much gnashing and googling I found that by removing /usr/bin/tls and re-running /sbin/ldconfig a few zillion times, I could make that go away although up2date still will not run in GUI mode, and thinks I'm on version 8.0, which apparently I am.

(3a) Because I can't get a v9.0 kernel running, I can't tell Red Hat Network that I'm wanting to subscribe to the Red Hat 9.0 channels, to which I have several active, paid-up entitlements! Therefore, even though I must presume that most of these problems have been fixed somewhere, I can't connect to RHN to download and install them.

---

(4) Until I hit upon the /usr/bin/tls kludge (3) above) I could not run mysqld nor mozilla nor even (now this is really serious! KPatience solitaire! Ordinary Red Hat tools like the services contr

(4a) Most if not all of the Red Hat stuff like redhat-config-services do not run:
Traceback (most recent call last):
File "/usr/share/redhat-config-services/serviceconf.py", line 35, in ?
import gnome.ui
ImportError: No module named ui

---

(5) Oh yeah, wine programs are stone-dead too. ("segmentation fault")

=== :cry:

Eleven hours later, Bangladesh has never called back; I presume they never will. ... :tisk: ... I have, apparently, a half-dead half-installed system that is quite cut-off from the RH9/Enterprise WS world that I was trying innocently to move to.

Where to start? I guess the first thing is to recommend a different way to upgrade. When you do a massive update in place it rarely works perfectly in spite of the vendors sales pitch. So you should set up some new partitions and install your new OS in parallel to the old. Then dual boot until your new system is stable. Then you can erase the old system. The dual boot method means that you don't go through a panic period where nothing works.

The same philosophy holds with massive updates using up2date. Most of the time things go smoothly but when they go wrong you are left trying to salvage a crippled system where you don't even know what up2date was trying to do when it screwed up. When that happens you have to abandon up2date and work through your problems using command line rpm.

All of what I have said so far may be irrelevent because you may be in too deep to restore back to your old system and try again.

If you are unable to reverse out of the mess then the first thing that I would do is to edit grub.conf and set yourself up with a dual boot of both the old and new kernel. Try to get as stable a system working as possible before you try any more massive infusions of up2date. It will take a few tests to determine if the old kernel or the new kernel is the most stable in the hybrid system you are stuck with.

Do you have backup to where you can restore back to your original system?

___________________________________
Be prepared. Create a LifeBoat CD.
http://users.rcn.com/srstites/LifeBo...home.page.html

Steve Stites