Red Hat Bugzilla – Bug 33949
Root RAID 1 causes obsolete MD ioctl error message
Last modified: 2007-04-18 12:32:27 EDT
From Bugzilla Helper: User-Agent: Mozilla/4.76 [en] (X11; U; Linux 2.4.2-0.1.28enterprise i686) I have been testing Red Hat Wolverine on my Dell PowerEdge 1300 server. Long story but I currently have everything running off a single 20Gb IDE drive transitioning towards three Ultra160 9Gb drives off an Adaptec U2W controller. I am moving that all to software RAID as soon as I can get it booting, which I can't currently. I created two soft RAID partions, /dev/md0 and /dev/md1. /dev/md0 is RAID 1 and /dev/md1 is RAID 5 and they both are reiserfs partitions. I used "find <old partition name> -xdev | cpio -pm <new partition name>" to move things from the IDE drive to the RAID partitions (which incidentally causes another problems - dmesg shows messages about running out of memory and the kernel starts killing processes, like httpd and mysqld but the copy finishes fine...). I setup lilo for a bootable RAID configuration and reboot. I haven't gotten the boot process to get past the LI prompt but I can boot from floppy disk. This works fine, the RAM disk loads, then the kernel image loads. It attempts to mount /dev/md0 and then the kernel panics with the following message: swapper (pid 1) used obsolete MD ioctl, upgrade your software to use new ictls <this is verbatim> kernel panic: unable to mount device (09:00) on VFS <from my flawed memory> The only problem, unless I'm mistaken, is no software, besides kernel software, is loaded yet. So what software needs upgraded? Is there a newer version of raidtools available? I've upgrade to kernel 2.4.2-0.1.28enterprise, raidtools-0.90-13, initscripts-5.78, lilo-21.4.4-13 and reiserfs-utils-3.x.0f-1 from RawHide after I was having these initial problems with Wolverine but nothing has made a difference. Reproducible: Always Steps to Reproduce: 1. Boot from root RAID 1, /dev/md0 Actual Results: As described, the kernel panics. The error message appears to be coming from linux/drivers/md/md.c, about line 2884, which appears to be the default if no other changes are specified for the RAID superblock. Expected Results: Should be able to boot from root RAID. Actually, one thing I haven't tried upgrading yet is SysVinit - I've just done that and will reboot with that change after I submit this. If it fixes the problem, I'll file a comment to this.
1) Reiserfs doesn't work on software raid and isn't reliable on hardware raid 2) Reiserfs is not supported by Red Hat and is included in the beta to asses the quality of reiserfs.
The other thing: We fixed several boot-raid bugs recently and we test this in our lab now; I assume this is fixed. Reiserfs recently got fixed to at least "work" with raid, although you still loose reliability.