On Fri, 2014-02-21 at 13:16 -0500, mulhern wrote:
Remove... is the same as before, but with extra parentheses.
Find more used...this was cherry-picked from master branch. It prevents the
previously observed problem (that a PartitionDevice shows up both in unused
and in new_devices, by leaving it out of unused).
But the effect of these two changes, which I think are both correct, is to
leave the accordion completely empty. Since there are no unused devices,
no Unknown OS/Unknown format page shows up. But since new_devices contains
no swaps and no mounts, the "New..." root doesn't get a page made for it,
either. And no pages in accordion leads to traceback.
So, I'm providing a new anaconda patch to go with this as well.
It's a wild guess.
We already talked about this, but I want to recap here for posterity.
The device that is causing this troublesome behavior is a newly created
device with only a DeviceFormat for formatting, which should never
happen in anaconda. The most likely origin for this device is a problem
in the patch to remove sibling lvm actions when hiding a disk --
specifically, since you are removing the actions in the same loop you
are detecting them in, I believe that at some point there is a PV that
doesn't get removed because it no longer meets the criteria in your
loop. If you split it out into a loop to build the list of actions and
another loop to cancel the actions you can avoid this problem.
David
David Lehman (1):
Find more used devices when calculating unused devices (#1043763)
mulhern (1):
Remove sibling physical and logical volume actions (#1043763)
blivet/__init__.py | 6 ++++--
blivet/devicetree.py | 4 +++-
2 files changed, 7 insertions(+), 3 deletions(-)