Used apt-get to upgrade to fedora. I had the same problems as joe and then some, read on... joe said: > Jay Daniels wrote: > >>I have been using yum to update packages and everything seems to work ok. >>Also noticed that some Fedora users use a port of apt-get. What is the >>"official" package updater for Fedora? Which works best apt-get port or >>yellow dog updater? >> >> >> > I mucked about with yum a bit and it didn't seem to be able to figure > out that it needed to use an http proxy, even though that was set in > /etc/profile, as well as yum.conf. apt-get worked like a charm though, > and I've been using it with great results - fills the role of up2date > from my RH8/9 days. I have used yum to get and install packages and yum automatically filled dependencies. I like yum, but your instructions on apt-get were pretty clear so I decided to upgrade my server with apt... > >>Someone posted that you can do a live update (dsl/cable) to fedora. Has >>anyone tried this and how? >> >> > Absolutely, I've remotely upgraded several RH8/9 boxes to fedora, and > they remained up and running, and in service, the whole time (It took > several hours, but as I said, I was logged into the boxes watching the > upgrade process the whole time) > > The basic idea is this: > > 1. install apt. > > 2. run the following commands to get your system up to date and ready > for upgrade > apt-get install > apt-get update > apt-get dist-upgrade > > 3. edit the /etc/apt/sources.conf to point to fedora core 1 repositories > instead of RH8/9 or whatever > > 4. repeat steps from #2 and watch as system is upgraded to fedora core 1 > > 5. I ran into a few snags where rpm complained during the upgrade about > file conflicts - in every case it was something to do with perl and > perl-cgi, etc. I removed the offending packages with rpm -evv --nodeps > and then continued with the upgrade by re-issuing the command "apt-get > dist-upgrade". Well, I broke perl and cgi but I did rpm -e offending packages ... and then ran apt-get dist-upgrade again and it worked, but I had changed to closer mirror and got errors because I did not have the correct gpg key for the rpms - the one on fedora.us didn't help. Once I imported the correct key located at fedora.redhat.com it worked. Took all night on dsl. > > 6. When all is done, the system will be basically upgraded, except for > the kernel, and will still be running on the old kernel with pretty much > everything else upgraded. At this point, you can say "apt-get install > kernel" to install the fedora core kernel, and at a time of your > choosing, you can then boot into the new kernel. The one thing I found I > had to install manually was fedora-release. "apt-get install > fedora-release" cleared that up. This is where I screwed up! apt-get update kernel-smp gave me a list of kernels and I didn't know anything about nptl, but I installed it anyway... > > 7. after you're up and running on the new kernel. you'll say "wow, the > debian fans had a point!" You can add a nice graphical update tool by > typing "apt-get install synaptic", and then begin pointing and clicking > your way to cool new programs of all sorts. > >>What if it fails? Linux box is gateway... >> >> > Depends on the nature of failure - possibly if you had too little disk > space it would fail partway through - but barring something like that or > loss of internet connection during the upgrade process, it should go > just fine. In any case, it deals with problems much more gracefully than > a normal install would. > > Hope this helps, > > Joe > I did not have disk space problems. I thought I would since my tmp is only 100mb partition, but apt-get put files in /var/apt/cache or something like that and I had plenty of space. $ df Filesystem 1K-blocks Used Available Use% Mounted on /dev/sda3 15116868 4096612 10252352 29% / /dev/sda1 295564 4690 275614 2% /boot /dev/sda2 74587868 2007480 68791460 3% /home /dev/sda8 4380004 569024 3588484 14% /home/web none 1034124 0 1034124 0% /dev/shm /dev/sda7 350021 1697 330253 1% /tmp /dev/sda5 10080488 1434140 8134280 15% /var $ df -i Filesystem Inodes IUsed IFree IUse% Mounted on /dev/sda3 1921984 210323 1711661 11% / /dev/sda1 76304 44 76260 1% /boot /dev/sda2 9486336 19903 9466433 1% /home /dev/sda8 557056 14 557042 1% /home/web none 258531 1 258530 1% /dev/shm /dev/sda7 90360 1623 88737 2% /tmp /dev/sda5 1281696 29294 1252402 3% /var However, I could not boot the new kernel and was forced to boot up with the old one. The new kernel got stuck on or after fixing /tmp and locked up. Doesn't make sense, but this is actually what happened... After the problem booting I booted from the old kernel. I must have ran rpm -e kernel-smp and removed the wrong kernel (my old one that I booted up with!) - then I realized my mistake and installed the kernel-smp rpms by downloading them using the lynx web browser. I also installed the correct kernel sources, etc. Anyway, after rebooting I was locked out of the system! Just after checking /tmp (successful or not) the kernel hung. Also had 3 broken soft links in /boot. I tried appending "linux 1" without the quotes using grub, but that didn't work. Tried my RedHat 9 cd "linux rescue" and it crashed with cpio errors (or something like that). I booted and tried to catch the "hit I for interactive..." but it boots so fast I couldn't get it to do interactive mode during kernel boot. Finally, after serveral attempts, it booted with the new kernel! I did startx and everything worked. RedHat update now uses yum, said there was a new kernel-smp and kernel-sources! So I let it install the newer kernel and sources. Rerunning apt-get install and apt-get update did nothing and showed no updates. The problem booting: I think the problem was this box has been up for months and I had some corrupted inodes or file (lot of rm -Rf) or I may have a hidden problem with this raid server, hint cpio problems suggest that I have io problems with the hardware correct??? You won't believe this but I installed the Microsoft XP Pro cd and the bastard booted to the new kernel! This box boots by default in this order, floppy, cdrom, hd. I checked the boot order several times, so how did it bypass the XP cd and boot up? I have to say I upgraded to fedora core 1, but as you can see I had many problems. I still can't understand the boot order thing above, but my system is now working ok and rebooted again to make sure. The rpm -e offending packages, broke a lot of stuff. I reinstalled the cgi and perl stuff and had to use apt-get to install mysql-server too for some reason. mysqld showed errors when starting, but I expect that's a bug. After rebooting and then restarting apache, everything works, squirrelmail, mysql, etc. Strange shit, I would hate to have do this again and don't recommend it to anyone unless you got a set of fedora disk on hand. RedHat and Fedora are the only 2 operating system I can get to install on this server out of the box. XP even requires special drivers for the 2120S raid controller and Debian Woody won't recognize the raid array or any disk. I tried the redhat driver from the supermicro cdrom, but Debian could not use it even after I put the drivers on floppy. Also Debian did not detect the built in ethernet devices. I wanted to use Debian, but upgrading to fedora was about my only option with this hardware: supermicro superserver, dual intel xeon... Still, doing a live update to fedora using apt-get was easier than downloading sarge and I really didn't know if sarge would work with my hardware either. $ uname -a Linux darkforce 2.4.22-1.2149.nptlsmp #1 SMP Wed Jan 7 12:51:51 EST 2004 i686 i686 i386 GNU/Linux Strange, i686 twice and i386. If anyone has RH 9 running like it should, I do not recommend upgrading to fedora at this time. Fedora seems bugger than RH 9 and X seems slower. If you must upgrade to fedora core 1, burn the cds first or wait until a more stable release of fedora. thanks joe! jay -- Shrike-list mailing list Shrike-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/shrike-list
