From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.3b) Gecko/20021225 Description of problem: I just finished an install of Phoebe on my second scsi drive, when trying to boot using floopy it hangs after it gets to RAMDISK: compressed image at block 0. This was an install over an existing Dell oem win2k setup. Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: 1.Install Phoebe on second scsi drive on Dell Precsion 330 2.Insert boot disk created during install in floppy 3. Try to boot off of floppy Actual Results: Hangs at RAMDISK: compressed image at block 0. Expected Results: Boot process completes Additional info:
I am able to boot using grub and the /boot from /dev/sda1. This is very simillar to the problem I started having with 7.3 and beyond. I was informed of an issue with recognizing ide drives and I have 2 cd roms in this machine along with 2 scsi drives. With 7.3 and 8.0 the installer would always hang on some post installation script, with Phoebe the install will complete, so progress has been made. With 7.2 I did not have this issue. Hope this helps, this box is available for any testing so feel free to ask for anything you want.
If you mount the floppy and compare the size of the initrd to the size of the initrd in /boot, is it smaller?
291831 Dec 28 02:36 initrd-2.4.20-2.2.img from /boot 253952 Dec 28 11:58 initrd.img frpm floppy
Added code in CVS to verify the sizes of the initrd + kernel written out to the floppy
*** Bug 78845 has been marked as a duplicate of this bug. ***
Does anaconda use mkbootdisk or some internal code? mkbootdisk returns an error if there is not enough space on the target media. The best solution would be to actually format fd0u1660 device if 1440 is not enough. There are cases where MBR is not accessible and user would be left with a non-bootable system with the only option: rescue mode, format fd0u1660 and mkbootdisk.
anaconda does use mkbootdisk. Currently anaconda detects when the bootdisk size is greater than 1440 and informs the user that it cannot create one. Experienced users can always use mkbootdisk themselves if they want to use unusual workarounds. The '--iso' option to mkbootdisk will create a bootable iso image that can be burned to a cd. Closing bug...
There have been several (un)releated problems and all got closed. The root cause I believe is that mkinitrd piles up all scsi modules it finds even though only one is needed on the image. I agree that experienced users will find a way to create a bootdisk. But anaconda is designed for unexperienced users and must desperately try to create a working bootdisk by a) removing extra scsi modules or b) formatting > 1440 floppy. This is definitely not a top priority for this release. I believe it should be documented. If system has two different scsi cards anaconda will most likely NOT create a boot disk.
It's too bad that removable media hardware is lagging, but anything that's done is just going to be a temporary bandaid. It's definitely not anaconda's place to do so. The anaconda side of this at least is fixed You can't even do boot disks on a lot of platforms these days, so the usefulness of it as a default is decreasing as time goes on.