Description of problem: We install our DELL2650 servers using kickstart via PXE. The installation kernel booted via PXE from network is images/pxeboot/vmlinuz with images/pxeboot/initrd.img from the original RH9 linux CD #1. The network kernel boot works fine, until anaconda comes into place throwing an exception. (I can post the full anacdump.txt). Note that this immediately happens after anaconda notes that there are not any partitions on the hard disk where Linux is going to be installed. Version-Release number of selected component (if applicable): Red Hat Linux 9. How reproducible: Always. Steps to Reproduce: 1. Install RH9 via kickstart and network boot (PXE). 2. Wait until anaconda tries to touch the hard disk. Actual results: Anaconda throws an exception. Expected results: Anaconda should install the system. Additional info: When booting from original RH9 CD instead of booting the kernel via PXE, we have no problem installing the system via kickstart completely using the same kickstart file.
I'm sorry, the platform is of course i686.
Please attach the actual dump, I have no data to work from without it.
Created attachment 97360 [details] anacdump.txt This anaconda's dump to diskette after throwing the exception.
Update: I found the part of the kickstart file that caused the exception: clearpart --all --drives=hda This is obviously wrong, since there is no hda drive on a DELL 2650, but an sda drive. But anaconda shouldn't throw an exception when the kickstart file tries do touch a hard disk that doesnot exist. So it is IMHO still an error that should be fixed.
Improving kickstart's error handling is an RFE somewhere. Right now, referencing drives that don't exist is a definite "don't do that" thing. The exceptions are better now than they used to be although distilling them and presenting them in a nice useful form to the user is still to occur.