Bug 1147845 - anaconda considers a full disk as completely empty
Summary: anaconda considers a full disk as completely empty
Keywords:
Status: CLOSED INSUFFICIENT_DATA
Alias: None
Product: Fedora
Classification: Fedora
Component: python-blivet
Version: 21
Hardware: Unspecified
OS: Unspecified
unspecified
unspecified
Target Milestone: ---
Assignee: David Lehman
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2014-09-30 08:15 UTC by Kamil Páral
Modified: 2014-10-06 16:45 UTC (History)
6 users (show)

Fixed In Version:
Clone Of:
Environment:
Last Closed: 2014-10-06 16:45:26 UTC
Type: Bug
Embargoed:


Attachments (Terms of Use)
anaconda.log (12.90 KB, text/plain)
2014-09-30 08:16 UTC, Kamil Páral
no flags Details
program.log (18.62 KB, text/plain)
2014-09-30 08:16 UTC, Kamil Páral
no flags Details
storage.log (133.68 KB, text/plain)
2014-09-30 08:16 UTC, Kamil Páral
no flags Details
system journal (364.58 KB, text/plain)
2014-09-30 08:17 UTC, Kamil Páral
no flags Details
rpm -qa output (46.08 KB, text/plain)
2014-09-30 08:17 UTC, Kamil Páral
no flags Details
fdisk output on that disk (622 bytes, text/plain)
2014-09-30 08:18 UTC, Kamil Páral
no flags Details
gdisk output on that disk (1.17 KB, text/plain)
2014-09-30 08:18 UTC, Kamil Páral
no flags Details

Description Kamil Páral 2014-09-30 08:15:20 UTC
Description of problem:
Yesterday I was helping a student with F21 installation. He had a completely full disk (Windows installation, 4 partitions, allegedly as provided by manufacturer and untouched), but anaconda displayed it a completely empty! That means the partition spoke overview, and also custom partition spoke - no existing partitions were shown.

That's a serious problem, because not only it would not be able to install alongside Windows, it would probably erase the disk contents completely, destroying all previous data.

I collected all the logs and hope it will help you debug the issue. I looked at the disk contents in several applications. gnome-disks showed the disk contents without any problems. gparted worked as well, but reported some warnings and errors (but honestly, I always see some warnings from gparted, so that doesn't need to mean much).

Version-Release number of selected component (if applicable):
F21 Alpha Workstation x86_64 Live
anaconda-21.48.6-1.fc21.x86_64
python-blivet-0.62-1.fc21.noarch

How reproducible:
always, for *that particular disk*

Steps to Reproduce:
1. check existing boot contents - boot windows (ok), display partitions in gnome-disks (ok)
2. run installer
3. see that anaconda reports roughly 590GB disk with 590GB free space, i.e. completely empty disk

Actual results:
the disk is reported as empty

Expected results:
the disk should be reported as full

Additional info:
F20 anaconda exhibited the same problem. Originally I wanted to install F20, then tried F21, the same problem.

Comment 1 Kamil Páral 2014-09-30 08:16:34 UTC
Created attachment 942620 [details]
anaconda.log

Comment 2 Kamil Páral 2014-09-30 08:16:37 UTC
Created attachment 942621 [details]
program.log

Comment 3 Kamil Páral 2014-09-30 08:16:40 UTC
Created attachment 942622 [details]
storage.log

Comment 4 Kamil Páral 2014-09-30 08:17:44 UTC
Created attachment 942623 [details]
system journal

Comment 5 Kamil Páral 2014-09-30 08:17:50 UTC
Created attachment 942624 [details]
rpm -qa output

Comment 6 Kamil Páral 2014-09-30 08:18:00 UTC
Created attachment 942625 [details]
fdisk output on that disk

Comment 7 Kamil Páral 2014-09-30 08:18:06 UTC
Created attachment 942626 [details]
gdisk output on that disk

Comment 8 David Lehman 2014-09-30 16:24:16 UTC
For some reason parted is refusing to instantiate a parted.Disk on sda. We sure don't give much information in these circumstances.

Comment 9 Brian Lane 2014-09-30 16:58:02 UTC
Can you run parted -s /dev/sda u s p on it? Since libparted is what is used that's the most useful output to see. If there's a bug in parted it should also show up when using the cmdline tool.

Comment 10 Brian Lane 2014-09-30 17:22:19 UTC
I'm not really sure what state this disk is in, hopefully you haven't let gdisk 'repair' it yet. I'd like to see the start and end of the disk:

dd if=/dev/sda of=first.64.sda.bin bs=512 count=64
dd if=/dev/sda of=last.128.sda.bin bs=512 skip=1250263600

Comment 11 Kamil Páral 2014-10-03 11:29:59 UTC
I'm sorry, I don't have access to that machine anymore. I hoped that those gathered logs and output would be enough. I forgot about parted, sorry. If we can't find the source of the problem from these logs, we will need to close this bug.

Comment 12 David Lehman 2014-10-06 16:45:26 UTC
python-blivet-0.61.4-1 will raise an exception if it finds a partition on a disk that parted says does not contain a disklabel.


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