lp:ubuntu/gutsy/devmapper

Related source package recipes

No recipes using this branch. (?)

Create packaging recipe

Related snap packages

Related bugs

Related blueprints

27. By Scott James Remnant (Canonical) on 2007-09-24

* Make the device take the /dev/mapper name in all cases.
  LP: #126379, #144049.
* Drop udeb patch since it's not needed.

26. By Colin Watson on 2007-07-16

* Disable the /dev/mapper/* -> /dev/dm-* symlink business in the udeb, as
  much of partman depends on the /dev/mapper/* names being canonical and
  gets confused by /dev/dm-*. For the time being, I've done this with a
  patch applied to the udeb's /etc/udev/rules.d/65-dmsetup.rules at build
  time rather than by maintaining two versions of the file, to avoid the
  rules getting out of sync.
* Fix syntax error in dmsetup(8) that broke the description of 'dmsetup
  info -c' under COMMANDS.

25. By Scott James Remnant (Canonical) on 2007-07-10

* Tweak udev rules to match discussion with SuSE. Note that having
  /dev/mapper/* as a symlink to /dev/dm-* is a temporary measure to make
  it easier to debug what's happening and will go away before release.
* Ignore devmapper devices with no name (often freshly created ones before
  a following change event).
* Change no_replace option to string_escape=none to match udev changes,
  adjust dependencies to match.

* Adjust devmapper node creation patch to avoid the loop; instead always
  rename over the top; also set the permissions and the selinux context
  on the temporary node so that we don't end up with half of udev's
  settings.

24. By Fabio Massimo Di Nitto on 2007-07-03

* Merge from debian unstable, remaining changes:
  - Ubuntu maintainer foobar
  - copy po/device-mapper.po to device-mapper.pot so Rosetta has a POT
    file to import
  - Demote dmsetup (and dmsetup-udeb) from Depends to a Recommends.
  - Apply patch from SuSE that adds an "export" option that will give udev
    the full device status of each dm device, including whether it's
    actually active or not.
  - Modify the udev rules we include.
  - Since both udev and devmapper might be creating the /dev/mapper device
    node, it's possible it can appear between our stat() and mknod() calls.
    So if mknod() returns EEXIST, do the stat again; also check lstat in
    case there's a hanging symlink there.
  - Install the udev rules in the udeb.
  - Add dmsetup postinst/preinst scripts.

23. By Scott James Remnant (Canonical) on 2007-05-29

* Don't run vol_id on devices with empty tables.
* But do run vol_id on devices with tables containing snapshot or
  snapshot-origin entries, since these are the final result, not the
  transient pieces.
* Also run vol_id on read-only devices. LP: #117089.
* Give devices with snapshot-origin in their tables a slightly higher
  link priority, so they win out of all the LVM devices (UUID will point to
  the base device, not any snapshots) but still not higher than the
  underlying devices so real hardware wins.

22. By Scott James Remnant (Canonical) on 2007-05-29

Further tweaks; always create the the /dev/mapper symlink and those in
/dev/disk/by-id; use the state and tables to only decide whether or not
to run vol_id, since that's the "problematic" bit.

21. By Scott James Remnant (Canonical) on 2007-05-29

* Fix a regression introduced by Kees's otherwise largely correct patch.
  We still need to not run vol_id on snapshot target devices, otherwise
  we'll end up using a changing device, and arguing over UUID/LABEL
  symlinks. (This might not be a complete fix, LVM may need to be patched
  to make inactive devices).

* Correct a school-boy error with the stat()/mknod() loop; we can't unlink
  then mknod() since that means there's a period without a device node,
  which could upset callers. Instead rename() the new device node over
  the top.

20. By Kees Cook on 2007-05-27

* Adjust debian/dmsetup.udev rule to not ignore "snapshot" devices. This
  will be needed even after "udev-lvm-mdadm-evms-gutsy" is solved (LP:
  #117225).
* Update Maintainer fields for Ubuntu.

19. By Scott James Remnant (Canonical) on 2007-05-21

Install the udev rules in the udeb, since these only arrange for
symlinks to point at the kernel name (replacing the /dev/mapper device
with a symlink), this isn't unsafe to do in the installer. LP: #105623.

18. By Scott James Remnant (Canonical) on 2007-05-21

* Merge from debian unstable, remaining changes:
  - copy po/device-mapper.po to device-mapper.pot so Rosetta has a POT
    file to import

* udev-lvm-mdadm-evms-gutsy:
  - Revert changes made during the feisty development cycle; while a
    valiant effort, we introduced just as many problems as we fixed.
  - Demote dmsetup (and dmsetup-udeb) from Depends to a Recommends, since
    we do tend to install those by default and there's no ordering problem
    anymore.
  - Apply patch from SuSE that adds an "export" option that will give udev
    the full device status of each dm device, including whether it's
    actually active or not.
  - Modify the udev rules we include; these now check the device state,
    set the /dev/mapper target as symlink to the kernel name (which HAL
    expects), and create the by-id, by-uuid and by-label devices itself.
  - Since both udev and devmapper might be creating the /dev/mapper device
    node, it's possible it can appear between our stat() and mknod() calls.
    So if mknod() returns EEXIST, do the stat again; also check lstat in
    case there's a hanging symlink there.