I'm thinking of making my root filesytem xfs as I've had it on another drive for a while and notice a real gain in performance under any highload over my ext3 root filesystem, so my question is is that if I can just simply do something like the page below describes http://www.blacksheepnetworks.com/se.../linux_xfs.htm
and just use my existing xfs filesystem to temporarily hold my root filesystem copy (without loosing any data from either my xfs or ext3 partitions!!) by doing something like "find / -mount -print | cpio -pdm /crap/temproot/" and then from knoppix make /dev/hda5 into xfs then just do that other command agian but kinda in reverse like "find /crap/temproot/ -mount -print | cpio -pdm /" (but only / will be whatever I have it mounted as on knoppix)


so heres my fstab:

/dev/hdb5 swap swap defaults 0 0
/dev/hda5 / ext3 defaults,noatime 1 1
/dev/hdb7 /transition vfat defaults 1 0
#/dev/cdrom /mnt/cdrom iso9660 noauto,owner,ro 0 0
/dev/fd0 /mnt/floppy auto noauto,owner 0 0
devpts /dev/pts devpts gid=5,mode=620 0 0
proc /proc proc defaults 0 0
#
/dev/hdb6 /crap xfs defaults,noatime 2 1
/dev/hda1 /windows ntfs nls=utf8,users,ro,umask=000 0 0
/dev/scd0 /mnt/scd0 iso9660 noauto,owner,ro 0 0

I'll make sure to make enough space for my / -> /crap/temproot/ to work but I just need to know ... well, will this plan of copying / to /crap/temproot then (from knoppix) /crap/temproot to whatever I have /dev/hda5 as work and I'll beable to boot back into slack9 and boot the kernel and NOT have a kernel panic about my root fs?
ps. xfs support is built into my kernels so no potential disaster there

Why XFS? Why not Reiser4? I have been doing some testing of my own and have found that XFS is great for huge files (It was developed by SGI for very large video files) Reiserfs (and the more recent reiser4) is much better for many small files along with the fact that it has superior journalling vs XFS.

Links
http://www.namesys.com/v4/fast_reiser4.html

I have tried reiserfs before (I think vs 3??) as it was in the wolk kernels that I use and the bare.i from slack9(I compiled wolk from src... and this was only a few weeks ago) yet whenever I put my disk that had reiserfs on it through ANY high i/o, I would lock up, since I had softdog on though, I could usualy get to a virtual terminal and manage a dmesg and it would tell me of a stack overflow in a program (different program each time such as xmms, gkrellm, mozillafirebird, or even gnome-terminal!!) but alas, my system was irrecoverable and would barely manage to go into init 0 (I couldn't even copy anything (1k txt files) without locking up). So after trying a few installations of reiserfs and horribly failing (trying to find what the problem was and forcing me to reinstall 2 times almost drove me nuts) I'm now very distrustfull of reiserfs, but xfs has held nice and stable although sometimes is a bit wierd with fs recovery when the kernel boots if something happened. Short story, I went back to ext3 (because I wanted to beable to turn my computer on) but I would rather like to use the speed of xfs so now I'm in a pinch.
ps. I have 2 WD drives, one 80 with the 8 mb cache and the other a 60 with a 2mb cache, I put them both through the manufacturers tests and they pulled through just fine, I have NEVER had any problems with anything besides reiserfs.

You should also checkout jfs. IBM is using it in all of it's "self healing" servers. I've never tried it but have heard it is great. After all, it is backed up by Big Blue!

Link
http://www-124.ibm.com/developerworks/oss/jfs/

well, yeah, xfs is jfs in a tweaked disguise! so its actually a bit better in some ways, xfs can perform directory listings and such under extremely high load like your computer was sitting idle which is quite a plus in my opinion as I often have to transfer or delete very large files, expecially if I've got an emergency and need to transfer them over to another drive fast when I don't have time