LIve CD persistent option seems to require casper-rw on USB (casper-cow)

[Richard Nemec]

Here's what I started with and then what eventually worked for me:
I had:
- Dapper LiveCD (tried Flight3, 02/14 and 02/16 daily)
- FAT formatted USB stick labelled 'casper-cow' with default Puppy Linux files in there.
- 'persistent' option added during the boot

This never worked for me - got into infinite script loop apparently trying to identify /dev/sda1 as casper-cow. Regularly stopped at 'sleep 5'.

After I renamed the USB stick under windows to CASPER-COW (upper-case, not sure whether it was needed), it seemed to get to a phase trying to find
/cow-backup/casper-rw. Again, looping with 'sleep 5'.

So I created an empty file 'casper-rw' on the stick - that got me past the visible mounting problems, but X never started.

Finally I noticed that it tried to mount (!) the casper-rw file as filesystem.
So I copied pup100 from Puppy Linux to casper-rw and everything started to work
(incl. persistence).

What I did is probably not the right solution but it may help:
- identify the source of my trouble
- help others to use my workaround until the boot script is fixed.

[Tollef Fog Heen (tfheen)]

I assume you have read https://wiki.ubuntu.com/LiveCDPersistence ?

Using FAT as the backing is not supported and there's no way it will work, please format the USB stick as ext3 or something similar.

The other option is to have a vfat file system on the USB stick and then put a file named casper-rw in there. Said file has to have a file system inside it, the easiest way to create that is to mount the USB stick, change to where it is mounted and run dd if=/dev/zero of=casper-rw bs=1M count=<how many mb you wish the file system to be> then run mkfs.ext3 casper-rw. It will then ask you to verify that you want to continue (since the file is not a block device), say yes.

Booting with the stick and "persistent" on the kernel command line should then just work.

I believe your problem stems from mixing up the "use my usb stick as a block device" and "store the persistent file system in a file on a vfat usb stick". Those don't mix.

[Richard Nemec (r-nemec)]

Hi, two things:
- thanks for fixing the wiki page
- the ext3 option still not working (see below)

Ext3 option:
Tried (again) the official wiki way - no success.
It seemed that it found the ext3 filesystem but then failed to use it.
So I decided to look for the script running this and found script named "casper".

Little poking around - and found exactly the place where it is IMO failing:

the function "find_cod_device()" when it finds the proper device (line 140)
by its vol_id, does
  echo "$part"
  return
Without knowing what exactly the script does, I'd assume it should say
  echo "$devname"
  return 0
instead.

[Tollef Fog Heen (tfheen)]

Richard Nemec wrote:

> the function "find_cod_device()" when it finds the proper device (line 140)
> by its vol_id, does
> echo "$part"
> return
> Without knowing what exactly the script does, I'd assume it should say
> echo "$devname"
> return 0
> instead.

Nice catch. Fixed in bzr, to be released in a little bit.

- tfheen

[Tollef Fog Heen (tfheen)]

casper (1.36) dapper; urgency=low

  [ Tollef Fog Heen ]
  * Fix typo in find_cow_device so block devices didn't work at all.
    Malone #31639. Thanks to Richard Nemec for the catch.
  * Update accessibility framework, thanks to Luke Yelavich.

  [ Colin Watson ]
  * Move the /cdrom mount into the new root filesystem.

 -- Tollef Fog Heen <email address hidden> Wed, 1 Mar 2006 11:13:54 +0100