Stuff that has not yet gone into the official build.
Post a reply

Re: Alternative usb installation method, part 3 (jessie)

Wed Jul 08, 2015 7:21 pm

have you ever run patch-live-initrd in a live session?

I have a vague memory of running out of space (ram and/or temp?). The unpacked initrd is around 200MB. Out of time just now to do any proper tests on this.

Re: Alternative usb installation method, part 3 (jessie)

Sun Jul 19, 2015 3:30 pm

Just did some more testing. Making notes here before I forget. Tested beta05 on a debootstrap devuan jessie amd64 install on hardware. (Most of the previous testing has been done in wheezy.)

Ran 'CRYPTSETUP=y update-initramfs -u' on host system before creating snapshot.iso.
Formatted the usb stick.
Added unpacked snapshot.iso
Created encrypted persistent partition.
Created encrypted persistent loopback file.
Created non-encrypted persistent loopback file.

All seems to be working correctly. Did not need to run update-initramfs during the live session. Did not use a patched initrd.

Note: Adding 'mountmode=rw,noatime' to the boot command allows root to write to live-media partition. Only tested this while booting with persistence, so I'm not sure if persistence is necessary.

Note2: Made a live-usb for someone last week, and both /lib/live/mount/medium and /lib/live/mount/persistence/<device> existed while running live. Not sure what that's about, but it could screw up some of the tests in the script. Have to look closer at that.

I think I'm near the point where the documentation can be updated, and it's ready to roll into a deb file.

Re: Alternative usb installation method, part 3 (jessie)

Sun Jul 26, 2015 11:29 pm

beta6 (in ibiblio testing)

Updated help file.

update-init-crypt.sh now works if live system uses findiso.

Re: Alternative usb installation method, part 3 (jessie)

Thu Aug 13, 2015 2:29 pm

Testing refracta2usb-2.0 with encrypted persistent volumes.

If you create an encrypted volume for persistence (partition or loopback file)
the script checks the initrd to see if it contains askpass. A warning is issued
if it's missing.


Made live-usb on jessie (mydevuan - devuan/angband/exegnu from a devuan debootstrap install)
Added the following live systems with the following results:


refracta-7.8 amd64
With persistent encrypted loopback file, got no warning
Booted with the entry for persistence. No password during boot, persistent volume not mounted, no r/w for user or root.
Added patched initrd. Got password prompt, persistence and root r/w media.
Added mountmode=rw,noatime and user gets r/w media.
(Note: got the same results with refracta-7.2 i386)


jessie-sysv i386
With persistent encrypted loopback file, with Warning
Booted persistence, got no password, persistent volume not mounted, but root has r/w media
Ran update-init-crypt.sh during the live session to update the initrd with cryptsetup.
Rebooted with updated initrd, got password prompt, persistence works, root has r/w media.


exegnu-refracta amd64
With persistent encrypted loopback file, with Warning
Booted persistence, no password, persistent volume not mounted, root has r/w media.
Added patched initrd. persistent volume not mounted. Sorry, I think my notes are wrong about who has r/w media, so that's not included here. In previous tests, I'm pretty sure r/w worked properly. Also, in previous tests I updated the initrd to get encrypted persistence to work.


mydevuan (added cryptsetup to initrd before creating the snapshot)
With persistent encrypted loopback file, got no warning.
Boot persistence, got password prompt, persistence and root r/w media.


Replaced the syslinux folder and mbr code with wheezy versions and retested persistence.
jessie-sysv still works with the updtaed initrd.
refracta-7 still works with the patched initrd.
mydevuan still works with the original initrd.

Looks like the preferred method (with isos made from jessie) is to add cryptsetup to the initrd before making the snapshot.

If you get the initrd warning with a wheezy-based iso, I think you'll need to patch the initrd, boot the persistence entry, update the initrd, boot back to the installed system and patch the updated initrd.

Anyway, I think it's about ready for prime time. Here a beta deb -
http://sourceforge.net/projects/refract ... b/download

Known or probable issues/bugs:
Some of the tasks end without a notification that says the task completed, which could be confusing at times.
Some of the messages about when you need to use a patched initrd might be wrong.
If you run patch-initrd after adding a live system, you'll need to Rescan, because the usb is still mounted. The error message now suggests a Rescan.
UNTESTED: Doing all this stuff while running in a live session (i.e. using a live-usb to create another live-usb or possibly adding another live system to the same usb.)

Re: Alternative usb installation method, part 3 (jessie)

Fri Aug 14, 2015 11:58 pm

Making a live usb from a running live usb works, but the menu entry is not right (in the beta). I made a copy of a running live system that was in /mydevuan on the source usb, and the menu entry on the target usb showed the path as /mydevuan/mydevuan. The actual directory was correct (/mydevuan, with kernel and initrd in /mydevuan/live/ instead of in /mydevuan/mydevuan/live as shown in the menu.) I've fixed it but haven't rolled a new deb yet.

Haven't tried making a live-usb from a running live-cd yet.
Haven't tried making a live-usb using findiso while running a live usb.

Re: Alternative usb installation method, part 3 (jessie)

Mon May 16, 2016 12:47 am

New live-boot 1:20160511 is out for Sid. (should be back-compatible and without other unstable deps).

Here is the updated patch-live-initrd : http://exegnulinux.net/refracta/experim ... 511.tar.gz .. Note you don't need to install anything to use it, it's for your "live" initrd only

The latest live-boot (or the patch) *may* solve the problem of refractasnapshot boot failure, which led to other "workaround" patches in previous versions and adding static nodes in /dev .. I can't verify that yet.

Re: Alternative usb installation method, part 3 (jessie)

Thu May 26, 2016 1:47 am

Updated refracta2usb to version 2.2.0 with the new files for patch-initrd. I also removed "union=aufs" from the boot line for ascii/stretch and beyond. It's not needed for jessie... unless you patch the initrd. I got dropped to busybox until I added "union=aufs" to the boot line. I'll add that to the boot options popup.

Re: Alternative usb installation method, part 3 (jessie)

Thu May 26, 2016 3:52 am

There was one download, and then I replaced the deb with a better one. Whoever it was should download again.

Fixed it so you can add union=aufs in a checkbox. Also, mountmode=rw,noatime wasn't getting automatically checked in one place (when adding a loopfile to first partition.) All seems to be working now.

Re: Alternative usb installation method, part 3 (jessie)

Tue Jun 14, 2016 5:08 pm

refracta2usb-2.3.0 is in the testing folder.
https://sourceforge.net/projects/refrac ... s/testing/

* Added support for UEFI.
* EFI files in iso get copied to usb.
* Boot menu entries get created for syslinux and grub-efi.
* Replaced archivemount with lsinitramfs in test for askpass.
* Removed/disabled support for zenity. (long ago broken by yad forms)
Post a reply