You are not logged in.
Hi Arch'ers,
For some years, I've been using the Clonezilla ISO to boot off from a USB-stick, for partition backups. But lately, I've been using my main harddisk partition as a full disk LUKS2-encrypted LVM-partition with detached header. But looks like I'm stopping with that from now on, because it doesn't work any longer (my setup is slightly different now, I think the detached header/LUKS2 is the culprit). I'm not completely sure what it is that makes Clonezilla fail to decrypt this LVM-partition - but my guess is that it first sees the partition (/dev/nvme0n1p6) and notices that it is marked as an LVM-partition and then it incorrectly tries to use the LVM-partition, before it's fully decrypted - but as it's an "invalid LVM-partition" it also shows up as "not used" (my guess)...
More concrete explanation:
When I come to the point where I have to mount the LUKS-encrypted partition and I write: "cryptsetup --verbose luksOpen /dev/nvme0n1p6 --header=..headerfile.bin --key-file=..keyfile.bin cryptLVM", Clonezilla fails with
"Cannot use device /dev/nvme0n1p6 which is in use (already mapped or mounted). Command failed with code -5 (device already exists or device is busy)."I tried "lsof /dev/nvme0n1p6" and "lsof | grep -i nvme" and several other things I can't remember (including combinations of "pvchange -a n" to try to "disable LVM" before lukssetup). The "lsof" prints absolutely nothing, so I can't understand/see what it's complaining about...
When I boot from the Arch Linux USB-stick and type in the EXACT same command, the LVM-partition becomes unlocked and I can access the LVM-partitions below the decrypted partition (root and swap). It's strange, I don't understand it. But (unless any of you know the explanation and maybe a solution), I've decided that now I want to use https://wiki.archlinux.org/index.php/Partclone and make future partition backups directly from the Arch Linux boot USB. The last few times I performed a clonezilla-backup, I think I did something like this:
(1) partclone.ext4 -c -s /dev/mapper/vg-root -o mybackupStorage/CLEAN_LINUX_NVIDIA_UNENCRPT/fresh_Arch_install.img
(2) /usr/sbin/ocs-sr -q2 -c -j2 -z1p -i 4096 -sfsck -senc -p choose savedisk 2018-12-19-00-img nvme0n1About the first command (1): I think this is called by the latter "ocs-sr"-script, for all individual partitions... If I only need a single partition, luckily I can directly use this command as I understand it. However, the latter command is the reason I wrote all of this, because it's a bit more confusing how I should proceed here:
About the latter command (2), I tried to see if I could run or install "ocs-sr" in Arch Linux, but that doesn't seem to be the case. I think
ocs-sris a wrapper, that does some clever things (found some part of it at https://github.com/stevenshiau/clonezil … bin/ocs-sr), so if I want to clone a full harddisk (all partitions) this "ocs-sr savedisk" command is run. But I don't understand it: It looks like it also uses e.g. "ocs-run-boot-param ocs_saveparts_postrun" and I might prefer a full partdisk-cloned image, which is better than just "dd" (more efficiently stored). The command automatically zips and -j2 apparantly means "--clone-hidden-data".
Question: How do I make something similar to what the "ocs-sr"-commandline (2) does in Arch Linux, if I boot from the Arch Linux USB and only have those tools available (suppose I want to create a fulldisk cloned image efficiently and don't seem to have the "ocs-sr" script available in arch linux, how to do the same and include MBR/GPT partition table etc)? Maybe the Wiki about partclone could be updated (https://wiki.archlinux.org/index.php/Partclone, I could even give it an attempt, but first I need to understand how to accomplish this like Clonezilla does it, efficiently) ? Looking forward to hear from you, thanks!
Offline
Hmm. Not much interest in this. Maybe I wrote a bad/boring explanation. I still think it would be a good idea to add a little bit to the wiki at https://wiki.archlinux.org/index.php/Partclone, about how to "replace ocs-sr" in an "Arch-linux"-way. So far, this is what I did:
fdisk -l > fdisk_partition_layout.txt
sfdisk -d /dev/nvme0n1 > sfdisk_partition_layout.txtAfter this, I think/believe that we can just do partclone.ext4, partclone.fat32, partclone.ntfs on all partitions... I think it's nice to have the partition info (start/end/size sectors), because that info I'm guessing isn't included in the partclone-data... And then I guess the "Microsoft reserved partition"-partition need to be fully cloned and handled separatly, e.g. using a tool like "dd", see e.g. partition 2 below?
Number Start End Size File system Name Flags
1 1049kB 274MB 273MB fat32 EFI system partition boot, esp
2 274MB 290MB 16.8MB Microsoft reserved partition msftres
3 290MB 135GB 134GB ntfs Basic data partition msftdata
5 135GB 135GB 537MB boot
6 135GB 1024GB 889GB ext4 lvm
4 1024GB 1024GB 547MB ntfs Basic data partition hidden, diagI wish there was something better than "dd" for partition 2, but anyway... Isn't it relevant to add to the wiki, how to get around using ocs-sr? Well, at least in my case I've been forced to do it (because clonezilla doesn't understand my LUKS2-encryption, I suppose)...? Anyone, have any good ideas or comments to add? Do you e.g. agree with my thinking or not? Thanks for any feedback/ideas/suggestion! :-)
Offline