Bug 651534 - Fedora 14 cannot be installed from NTFS local hard disk while Fedora 13/12 could
Summary: Fedora 14 cannot be installed from NTFS local hard disk while Fedora 13/12 could
Keywords:
Status: CLOSED WONTFIX
Alias: None
Product: Fedora
Classification: Fedora
Component: anaconda
Version: rawhide
Hardware: Unspecified
OS: Linux
low
medium
Target Milestone: ---
Assignee: Ales Kozumplik
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2010-11-09 18:53 UTC by Rajeesh
Modified: 2014-09-30 23:39 UTC (History)
8 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed: 2011-03-03 14:45:18 UTC
Type: ---
Embargoed:


Attachments (Terms of Use)
Anaconda log, from tty3 (1.85 KB, text/plain)
2010-11-09 18:53 UTC, Rajeesh
no flags Details
program.log (86 bytes, text/plain)
2010-11-09 18:54 UTC, Rajeesh
no flags Details
syslog, from tty4 (80.79 KB, text/plain)
2010-11-09 18:55 UTC, Rajeesh
no flags Details

Description Rajeesh 2010-11-09 18:53:53 UTC
Created attachment 459215 [details]
Anaconda log, from tty3

Description of problem:
Though installation notes does not mention NTFS filesystem is supported for installation from local hard disk, it used to work with F13, F12 and possibly F11. In F14, anaconda is stuck at STAGE2 looking for install.img

Version-Release number of selected component (if applicable):
14.22-1.fc14

How reproducible:
Always

Steps to Reproduce:
1. Follow steps for "Installing without media"
2. Keep the "images/" directory with contents and Fedora-14-x86_64-DVD.iso in an NTFS partition
3. Choose "Install from Hard Disk" and give correct path to the NTFS device
  
Actual results:
Installation is stuck, with no progress. Only way out is ALT+CTRL+DELETE

Expected results:
Installation proceeds by accessing "images/install.img"

Additional info:
The boot parameters passed are: "asknetwork debug=1 loglevel=debug noselinux syslog=<host>:<port>"
Looks like mount never returns. Attaching anaconda.log, syslog and program.log captured by remote logging.

Comment 1 Rajeesh 2010-11-09 18:54:48 UTC
Created attachment 459216 [details]
program.log

Comment 2 Rajeesh 2010-11-09 18:55:48 UTC
Created attachment 459217 [details]
syslog, from tty4

Comment 3 Cory Bell 2010-11-11 08:06:26 UTC
This happens to me as well, it seems to hang mounting the NTFS partition, doesn't matter whether it's stage2 or the repo. For instance, if I move stage2 (the install.img) to a FAT32 partition, then I'll get into the installer, but hang as soon as it mounts the NTFS partition to access the repo. If I leave stage2 on the NTFS partition, then it hangs just after the language selection when it tries to access stage2. I've tried manually specifying the location, (i.e. specifying /dev/sdXX and the path during interactive install), and I've tried using the boot command line options to find it automatically like so:
repo=hd:LABEL="Portable_HD":/iso/Fedora-14-i386-DVD
or
stage2=hd:LABEL="Rescue_Boot":/boot/images/fedora14_i386 repo=hd:LABEL="Portable_HD":/iso/Fedora-14-i386-DVD

Both of these volumes are on the same USB external hard drive, Rescue_Boot is a FAT32 partition on /dev/sdX1 (booted via GRUB), and Portable_HD is an NTFS partition on /dev/sdX2

I can get a log of the second scenario (stage2 on FAT32, repo on NTFS) if that will help, but it looks exactly the same as Rajeesh's output - the attempted NTFS mount and then nothing.

This configuration worked perfectly for me in FC12 and FC13, not certain about FC11.

Comment 4 Ales Kozumplik 2011-01-11 18:17:09 UTC
Tom,

Anaconda hangs whenever we try to mount a ntfs volume because our mount wrapper always waits for the mount command to terminate so we can log the output. I talked to rjones and he told me mount.ntfs doesn't exit since it's a FUSE type filesystem.

Could it at least close stdout and stderr so we know we shouldn't wait for more output from it? (In that case feel free to reassign this to yourself)

Thanks.
Ales

Comment 5 Tom "spot" Callaway 2011-01-12 17:38:32 UTC
I did a scratch build that simply adds 

fclose(stdout);
fclose(stderr);

to the end of main() in mount.ntfs. I'm not sure this will work right though, if fuse is doing anything after mount.ntfs is done.

http://koji.fedoraproject.org/koji/taskinfo?taskID=2717362

Please test it and let me know.

Comment 6 Ales Kozumplik 2011-01-13 09:14:05 UTC
No, that didn't help. Which means we have a general problem with fuse. Peter, koji says you are the fuse maintainer, can you help? Is there a reason to keep the stdout and stderr opened after mount.ntfs-3g goes to background?

Thanks.
Ales

Comment 7 Ales Kozumplik 2011-01-25 11:34:32 UTC
Today I contacted the fuse maintainer directly. If noone can provide the needed information I am going to have to remove support for mounting ntfs (and other fuse filesystems).

Comment 8 Tom "spot" Callaway 2011-01-25 16:06:28 UTC
Or, you could modify your mount wrapper to timeout after a certain period of inactivity.

Comment 9 Ales Kozumplik 2011-03-03 14:45:18 UTC
I am closing this: starting with F15 and onwards there's no stage2 image to be mounted from a ntfs.

Comment 10 Hedayat Vatankhah 2011-10-22 22:40:43 UTC
But there is Fedora DVD iso image which should be mounted from an NTFS partition so that we can install Fedora from an NTFS partition...


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