[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: mdadm device assembly (was: udev causing data loss?)



also sprach Boyd Stephen Smith Jr. <bss03@volumehost.net> [2008.11.15.2020 +0100]:
> Are mdadm's UUIDs not the same as udev's?

mdadm in lenny has udev rules to create links under /dev/disks, if
that's what you mean. But the kernel does not (yet) generate udev
events when md arrays are added or removed.

> That seems sane, not twisting at all, to me.  Last I checked this
> is also the way LVM works.  Will lvm2 be getting a similar "fix"?

lvm2 is in much the same boat.

Take

  mdadm --assemble --auto=yes /dev/md0 /dev/sd[abc]1

This creates the /dev/md0 device node and then uses ioctl()'s to
assemble the array behind it, using the components in the first
partition of the drives.

A more natural way would be to tell the kernel to assemble the
components via /sys, and then, once the device is ready, give it
a name (which could come from persistent data in the superblock),
and have something create the appropriate node under /dev.

But as said, mdadm will likely grow the second capability in
addition, not instead of the first.

-- 
 .''`.   martin f. krafft <madduck@debian.org>
: :'  :  proud Debian developer, author, administrator, and user
`. `'`   http://people.debian.org/~madduck - http://debiansystem.info
  `-  Debian - when you have better things to do than fixing systems
 
because light travels faster than sound,
some people appear to be intelligent,
until you hear them speak.

Attachment: digital_signature_gpg.asc
Description: Digital signature (see http://martin-krafft.net/gpg/)


Reply to: