From Bugzilla Helper: User-Agent: Mozilla/4.73C-SGI [en] (X11; U; IRIX64 6.5 IP30) I decided to perform a clean install over a previous Redhat 7.0 Linux install. The harddisk is a 45 GB IBM IDE drive, with a 30 GB windows partition at the beginning and the remaining 15 GB used by Redhat 7.0 (the automatic partitioning option of RH 7.0 for the workstation install was used on this disk). If I try a clean RH 7.1 install on this system, the automatic partitioning option does not appear, even though it worked just fine with this arrangement in 7.0. Disk Druid also does not allow me to create a /boot partition because it will lie > 1024 cylinders. Again this was no problem in RH 7.0. Lilo worked fine in RH 7.0 with the boot partition created by the automatic partitioner in that version, as long as I added lba32 to the lilo.conf created by the RH 7.0 installer. It seems that the new disk druid / automatic partitioner does not like partitions above 1024 cylinders even though the work fine with RH 7.0. I have tried several time, with the non dos portion of the disk undisturbed from the previous RH 7.0 installion, completely unpartitioned, partitioned as a single DOS partition, and partitioned as a single ext2 partition -- in every case both the automatic partitioner and disk druid fail. Reproducible: Always Steps to Reproduce: 1. partition a disk with a large dos partition and empty, unpartitioned space after 1024 cylinders 2. try to install RH 7.1 3. Actual Results: Disk Druid fails Expected Results: The boot partition should have been created, as it was in RH 7.0
What do you mean by 'fails'?
By fails, I mean that disk druid will not allow me to create a /boot partition, as it keeps issueing the > 1024 cylinder error message. I could not find a way to manually partition the disk that would allow me to continue with the RH 7.1 installation process. Basically, I could not get past the partitioning screens. I finally gave up, put in the RH 7.0 CD, rebooted, and it installed without any problems.
Red Hat Linux 7.0 did not create /boot above the 1024 limit automatically. If it did then this was in error. We never wrote the lba32 option out in the 7.0 installer because we decided the option was too experimental in our experience. In 7.1 we made a slightly different attempt. If you manually partition your drive with fdisk, then assign '/boot' to a partition above the 1024 limit, you should see a message saying to make a boot disk in case the lba32 option does not work. Then when we write out lilo.conf we'll put the lba32 option in the file. So try partitioning manually and see if this works better.
I would claim that the RH 7.0 error is a desired feature then. With RH 7.0, the automatic partitioner creates my /boot and / partitions both above 1024 cylinders. The disk druid in RH 7.1 confirmed that the partitions created by RH 7.0 were all above 1024 cylinders. I did try doing the manual partitioning with disk druid in RH 7.1, but if I understand your reply correctly, the installation procedure will only proceed if I use fdisk to create the partitions-- is that correct? I installed RH 7.0 on my system and I will most likely leave it there for a while, as I do not want to go through the trouble of a reinstall again. Unfortunately an upgrade to RH 7.1 also fails on this system for a different reason, but that's another story. I will most likely wait until the next .1 release before trying again. I have read atleast one other report of this problem posted in the question area of linuxtoday.com. Mike
The other reference to this problem that I referred to above is at: http://losc.internet.com//read.php?f=13&i=1845&t=1845 Mike
That's right. Does the install proceed correctly if you use fdisk to make a /boot above 1024?
Closing due to inactivity. Please reopen if you have more information.