Package: util-linux
Severity: wishlist

Please add support for detecting RAID array membership and skip
attempting to read partition tables of disks that are members of an
array. Or, if already doing so for linux software RAID, please add
support for fakeraid.

As you can read in the discussion that took place under bug #778712, a
few days ago I was attempting to write a GPT partition table to a
fakeraid RAID0 array that I had just setup using my motherboard firmware
(not knowing at the time that linux software RAID is better supported
and recommended unless dual booting with Windows).

This turned into a confusing mess leaving me thinking that parted might
have a grave bug. Actually it was just a combination of:
1) parted not understanding RAID array membership, resulting in
confusing info being output.
2) fdisk also not understanding RAID array membership, resulting in
confusing info and an error being ouput.
3) fdisk (understandably) not flushing the disk caches of the member
disks after writing a partition table to the array, thus leading me to
think that doing 'parted -l' afterwards was modifying the disk, breaking
things, when in fact all it was doing was flushing the caches, allowing
the true state of things to be revealed in the next use of 'fdisk -l'.
4) My having no clue that any caches were involved in any of this.
5) The (default) stripe size used for the array being 16KiB, small
enough for the GPT table to be split across the two disks, resulting in
it appearing corrupt when the tools viewed an array member as a
standalone disk.

If fdisk flushed every disk cache when performing 'fdisk -l', like
parted does, then I never would have attributed a problem directly to
parted, however I am not advocating such a change, since if only fdisk
and parted understood array membership, then no confusion would have
resulted whatsoever. Please add such support.


-- 
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org

Reply via email to