From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.9) Gecko/20020408 Description of problem: System info: gBox P4 (CF-S868) Flex-ATX (http://www.amselectronics.com) 512MB RAM 2x 80GB Western Digital Drives (WDC800JB) Problem: Anaconda crashed whenever you have at least one raid (auto) partition on any of the drives. The crash occurs at line 43 of the /usr/lib/anaconda/raid.py file right after it finished formating all the partitions. My first try was a simple install using 2x250MB on a raid1 for /boot, 2x10GB unused partitions (intended for windows dual boot, but initially set to unsused raid), and 2x60GB+ on a raid0 for /. This failed, so I tried with an ext3 /boot on just one of the 250MB partitions and the rest the same; failed. Next I tried eliminating the raids from the install by putting /boot on one of the 250MB partitions, and / on one of the 10GB partitions, both on hda, failed again. Next we tried the same scenario, but deleting all the partitions of type raid; it worked. So, my take on it is that it fails while scanning for autodetect raid partitions. Logs: anaconda.txt is attached Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: Additional info:
Created attachment 71184 [details] traceback dump of anaconda state generated by itself after crash
If you have any partitions set as type raid they need to be part of a valid raid array.
And you think this is ok behaviour? I mean, what if I have 5 or 10 disks, and I want to pre-partition as raid autodetect partitions to be used later? In my case I had a 10GB partition on each drive that I was going to use later on as a mirror or stripe, but first I was going to use one of them to install windows and do some testing, so I defined them as type "fd". Unless we are know in competition with Microsoft for bad programming, this certainly bad behaviour. At the least, if you do not want the user to define raid partitions that are not used, then give out a message, but crashing the whole install program! duh!
Yes it is certainly something we would like to address. We are getting to these types of issues as time allows. *** This bug has been marked as a duplicate of 32397 ***