Bug 496363 - anaconda refuses to mount encrypted file systems
Summary: anaconda refuses to mount encrypted file systems
Keywords:
Status: CLOSED RAWHIDE
Alias: None
Product: Fedora
Classification: Fedora
Component: anaconda
Version: rawhide
Hardware: All
OS: Linux
low
medium
Target Milestone: ---
Assignee: David Lehman
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2009-04-18 04:35 UTC by Clyde E. Kunkel
Modified: 2009-05-19 20:20 UTC (History)
5 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed: 2009-05-19 20:20:49 UTC
Type: ---
Embargoed:


Attachments (Terms of Use)

Description Clyde E. Kunkel 2009-04-18 04:35:58 UTC
Description of problem:
During clean install of rawhide 20090417 using custom partitioning, specified that some encryted LVs be mounted on /mnt/...., but during first boot passwords were not prompted, LVs not mounted and directories in /mnt did not exist

Version-Release number of selected component (if applicable):
11.5.0.45

How reproducible:
each time

Steps to Reproduce:
1. boot.iso, askmethod
2. select custom partitioning
3. edit encrypted LVs requesting mount on /mnt/...
  
Actual results:
Looked good, but on first boot and subsequent, they were mot mounted, nor did directories exist in /mnt

Expected results:
mounted encrypted partitions (LVs)

Additional info:

Comment 1 Clyde E. Kunkel 2009-04-19 21:08:29 UTC
In 11.5.0.46, you must first custom mount another LV in a VG, then custom mount the encrypted LV, otherwise anaconda ignores the mount point you specify.  This may have been the case in .45, I didn't think to revisit a VG with that version.

Comment 2 Mike Wolf 2009-04-21 02:39:25 UTC
I think I'm experiencing the same bug, (rawhide 20090420) though through a slightly different method.  I had a working installation on a test machine where the VG on /dev/sda2 was encrypted, and I had created another encrypted ext4 partition directly at /dev/sda3 which was mounted as /home.  Then I reinstalled...

I was able to remount everything except the encrypted ext4 partition on /dev/sda3.  This was the only partition/volume that I chose not to format, since I wouldn't be formatting that partition if it were a system I used every day.

I later tried recreating this scenario without any encrypted partitions.  Without encryption, /dev/sda3 mounted to /home with no problem.

Comment 3 Clyde E. Kunkel 2009-04-22 22:19:12 UTC
pre-existing encrypted partitions still not mountable in anaconda 11.5.0.47 unless you first edit another LV in the VG.

Comment 4 Clyde E. Kunkel 2009-05-03 18:55:33 UTC
You still must edit an encrypted file system two times in a row to get anaconda to record the mount point.  .48 version.

Comment 5 Clyde E. Kunkel 2009-05-08 15:44:50 UTC
Still a problem in version .50.  Also, 499828 may have some relationship.

Comment 6 David Lehman 2009-05-08 16:00:42 UTC
(In reply to comment #3)
> pre-existing encrypted partitions still not mountable in anaconda 11.5.0.47
> unless you first edit another LV in the VG.  

Have you seen this with partitions also or only with LVs?

Comment 8 Clyde E. Kunkel 2009-05-08 18:42:34 UTC
Not with non-encypted partitions, I only have encrypted LVs.  Give me a couple hours and I'll build a test case.

Comment 9 David Lehman 2009-05-08 18:59:25 UTC
Don't spend the time unless you want to -- encrypted partitions have been addressed already and should work just fine. It was your use of the word "partitions" in comment #3 that made me wonder if you were seeing it with partitions or just using terms loosely.

Comment 10 Clyde E. Kunkel 2009-05-08 19:41:32 UTC
Didn't see #9 in time, so went ahead and built the test caser and encrypted partitions edit the first time.

I will be more careful about describing "partitions" and "LVs."  Thanks for pointing this out.

Comment 11 David Lehman 2009-05-11 18:55:57 UTC
Fix will be in anaconda-11.5.0.52-1. Thanks for the report.

Comment 12 Clyde E. Kunkel 2009-05-14 18:23:41 UTC
with anaconda .52, the mount point stays in the dialogue after hitting ok, but then after hitting the next ok to go back to the custom partitioning screen, traceback: "luks object has no attribute."  Auto bz said 499828 had already been submitted, traceback is there.

Comment 13 Clyde E. Kunkel 2009-05-19 20:20:49 UTC
Working in .53.  Closing.


Note You need to log in before you can comment on or make changes to this bug.