[blivet][master/rhel7-branch] clean up hiding/unhiding of disks

Vratislav Podzimek vpodzime at redhat.com
Thu Sep 18 07:10:47 UTC 2014

On Tue, 2014-09-16 at 16:01 -0500, David Lehman wrote:
> The first patch is mostly unrelated, but here it is anyway. It is not
> intended for rhel7-branch at this time, but the others are.
> The other four are all related to hide/unhide as used by anaconda for
> disk selection. They tighten up management of relationships between
> devices and the devices that contain them when adding and removing
> devices from the tree, whether it be part of registering an action or
> hiding a disk. The biggest piece is patch 3, which restores relationships
> between lv and vg, or partition and disk, when canceling actions that
> removed the lv or partition. Patch 5 is especially nice now that we are
> hiding zram devices on every entry to the custom spoke -- without these
> patches your previous custom layout would be wiped out upon entry to the
> custom spoke since hiding the zram device would cancel all registered
> actions.
> David Lehman (5):
>   Do not allow modification or removal of protected devices.
>   The first format destroy action should obsolete any others.
>   Improve handling of device removals/additions from the devicetree.
>   Cancel actions before hiding descendent devices.
>   Only cancel actions on disks related to the one we are hiding.
>  blivet/__init__.py     |  11 ++++-
>  blivet/deviceaction.py |  27 +++++++++----
>  blivet/devices.py      | 108 +++++++++++++++++++++++++++++++++++++++++++++++++
>  blivet/devicetree.py   |  70 ++++++++++++++------------------
>  blivet/partitioning.py |   6 +--
>  tests/action_test.py   |   6 ++-
>  6 files changed, 176 insertions(+), 52 deletions(-)
I'd like to say big THANKS and big ACK for those patches as they clean
things up a lot. I've been debugging issues with device (un)hiding and
actions cancellation for many days last week and it was hell. With
these changes, any additional fixes (if needed) should be much more
straightforward and clear.

Just have a look at the neatpick comment on PATCH 5/5.

Vratislav Podzimek

Anaconda Rider | Red Hat, Inc. | Brno - Czech Republic

More information about the anaconda-patches mailing list