The xfsprogs are too old to create the filesystem on software raid devices. They will exit with the error: "MD array not in clean state" Even though they are in clean state. I was trying with the livecds install-x86-2004.2-r1-minimal.iso and install-x86-minimal-2004.2.iso - I just mention this because somebody else could have this kind of trouble. I assume that 2004.3 livecds will have a newer version of xfsprogs with them. I'd say the bug can be closed when 2004.3 is out and somebody successfully executed mkfs.xfs /dev/md0 I managed to install gentoo on SW-RAID1 and XFS using this rescueCD: http://www.sysresccd.org/ Reproducible: Always Steps to Reproduce:
Is ther version of xfsprogs that is marked stable for your arch just not up-to-date? We do not modify ebuilds for the LiveCD. If it is not up-to-date in the portage tree, then it won't be up-to-date on the LiveCD. If the package in portag eis not up-to-date, then I would suggest filing a separate bug and assigning it to the package maintainer, then marking this bug as a dependency of that bug. If the maintainer does nothing, I can put some pressure on them to get it done, but otherwise I can't do much of anything without their permission. Also, this has to be completed by the 20th of October or it won't make it on 2004.3 at all.
There is sys-fs/xfsprogs-2.6.13 stable which works with sw-raid, seems like the 2004.2 release has xfsprogs 2.3.x which was old even at release-time (i have found some reports of gentoo users at sgi's bugzilla). The people reporting the bug made gentoos xfsprogs maintainers fix it, and their problem was gone (just emerge -U it with custom ebuild) - mine was not, because upgrading while running the livecd for fresh install is impossible. I just posted this to give others a help-guide and to remind the livecd makers that xfsprogs should be a recent version for the next release. Which it is going to be, because the current stable version will be used according to your post. I don't know when to declare this bug as fixed - when 2004.3 is out and somebody tried it - when somebody states that the current stable version should work (a.k.a. right now)?
I am building the test LiveCD's for 2004.3 right now, so I would say since we're using the newer versions on the livecd now, that this is resolved now... Sound good?