[SOLVED] Cloning the SSD to an nvme

[tired spire]

I cloned my Arch Linux system from a 223GB SATA SSD to a 1TB NVMe using "dd" (full disk clone, including partitions).

After cloning, I resized the partition on the NVMe to use the full disk space. The partition resize worked fine, but the filesystem didn’t expand.

My root filesystem is btrfs.

When I boot from the NVMe, the system works, but the root filesystem is mounted as read-only, and I get errors like:

• "parent transid verify failed"
• filesystem forced into read-only mode
• cannot resize: read-only filesystem

"dmesg" shows multiple btrfs errors related to transid verify failures.

Important detail:
Both the original SSD and the cloned NVMe are connected at the same time, and since I used "dd", they have identical filesystem UUIDs.

Current state:

• System boots from NVMe
• Filesystem is read-only
• Resize via "btrfs filesystem resize max /" fails due to read-only state

What I suspect:
Btrfs gets confused because both drives have the same UUID and are present at the same time

What I want:

• Fix the read-only issue
• properly expand the filesystem to full 1TB
• keep the old SSD as backup if possible

What would be the safest way to fix this without risking data loss?

My old System was the SSD, the new one should be on the 1TB nvme.

Thanks for helping!

(More infos follow once I can use discord on that damn system)

https://pastebin.com/KDAeXZzb

[tired spire]

Changed uuid of the SSD and now nvme doesn't boot
I did that because I think nvme-linux uses the root subvol of SSD-linux

please someone help me I'm kinda worried I can't use that pc this weekend and I kinda need it

[lean saffron]

take the old ssd out first

do resize and change uuid

and setup boot etc etc

and then plug it back in

and you should be good

[tired spire]

I can't take it out

It's really hard to do that

[lean saffron]

okay uhh

you already changed uuid right?

[tired spire]

Also I don't know how to setup these partitions

[tired spire]

Yes from.live usb

[lean saffron]

there are two uuids you need to change

partition uuid and filesystem uuid

after that just setup boot to use the new UUIDs and you should be good

[tired spire]

sudo btrfstune -u /dev/sda2
I think it was that here

Sda is my ssd

[tired spire]

I changed both of them and edited fstab
But btrfs says 20000 filesystem errors found

[lean saffron]

uhh

not corrupted clone?

[tired spire]

I think yes

Wait let me look

I still got the CatchyISO

[dry plover]

boot the cachy iso, and do not mount either drive. we need to first check the integrity of btrfs on SSD (sda2?)

[tired spire]

Okay I'm already booting

[dry plover]

show an lsblk -f after booting

[dry plover]

good they're both unmounted, and original ssd is sda2 right?

[tired spire]

Yes it's 200G so it must be the original one

[dry plover]

run sudo btrfs check --readonly /dev/sda2

[dry plover]

ok hold on

[tired spire]

Is that bad?

[dry plover]

I think it just means the btrfstune from earlier was interrupted

one sec I'm getting the right command from man page

[tired spire]

Okay thanks

[dry plover]

sudo btrfstune -u /dev/sda2

[dry plover]

sorry the picture is kinda unclear, did that command finish with no errors?

[tired spire]

Yes no errors

[dry plover]

good

now sanity check sudo btrfs check --readonly /dev/sda2

[tired spire]

Same as before...

If we can't copy the files from SSD to nvme I could as well also install catchyos again and copy everything it would just take a few days but as we can't get to the SSD data yet... that probably isn't an option

[dry plover]

show the error this time

[dry plover]

weird... fsid change finished, and then, it didn't actually finish?

I'm researching

output of sudo btrfs inspect-internal dump-super -f /dev/sda2 | grep -iE 'fsid|metadata_uuid|incompat|generation' ?

[tired spire]

I'm having issues with that en keyboard layout too but loadkeys de doesn't seem to work?

[dry plover]

sudo btrfs inspect-internal dump-super -f /dev/sda2 | grep -i log_root

[dry plover]

here's an explanation of this (from claude but it's accurate in this case): "the check error specifically says "Couldn't setup log root tree". The bad block at offset 66781184 is the tree-log root, which btrfstune deliberately doesn't rewrite because the tree-log is supposed to be transient (fsync-replay data that gets consumed and discarded on next clean mount). The SSD was uncleanly shut down at some point during this saga, so there's an unreplayed log root pointer in the superblock pointing to a block written before the fsid change. That block legitimately still has the old fsid 39f28f6e-… because btrfstune skipped it, and now check chokes when trying to validate the log root. The rest of the filesystem is actually fine"

we don't need to fix that though since sda2 should be able to be mounted read only at this point, and then we can rsync it to the nvme

sudo mkdir -p /mnt/ssd

[tired spire]

Done

[dry plover]

now do sudo mount -o ro,nologreplay,subvolid=5 /dev/sda2 /mnt/ssd

[tired spire]

Thanks, atleast something I can understand 😂

[static gust]

alpha.lynx received a thank you cookie!

[tired spire]

unknown parameter nologreplay

[dry plover]

oops, do this instead sudo mount -o ro,rescue=nologreplay,subvolid=5 /dev/sda2 /mnt/ssd

[tired spire]

Yep done

[dry plover]

now ls -al /mnt/ssd

[dry plover]

okay, good ssd has the data

now show another lsblk -f just to make sure nvme is unmounted

[tired spire]

Yep it's unmounted

[dry plover]

sudo wipefs --all /dev/nvme0n1p1 /dev/nvme0n1p2 /dev/nvme0n1

[tired spire]

Sorry if it's unclear but no errors
(Can't unplug my phone every half second)

[dry plover]

now create fresh partitions on the nvme, sudo cfdisk /dev/nvme0n1

[tired spire]

Label type?

[dry plover]

select new GPT layout, and create partition 1 with 512 MB, change it's label to EFI System

and partition 2 the rest of the space

[tired spire]

Write?

[dry plover]

yes

write then quit

[tired spire]

Done

[dry plover]

just to verify do lsblk -f

[dry plover]

make the fat32 fs sudo mkfs.fat -F 32 /dev/nvme0n1p1

and btrfs sudo mkfs.btrfs /dev/nvme0n1p2

[tired spire]

Done with no errors

[dry plover]

sudo mkdir -p /mnt/nvme

sudo mount -o subvolid=5 /dev/nvme0n1p2 /mnt/nvme

[tired spire]

Both done

[dry plover]

create the subvols, for sv in @ @home @root @srv @cache @tmp @log; do sudo btrfs subvolume create /mnt/nvme/$sv; done

[tired spire]

How do I load my keyboard layout

de layout

Because I can't find ; in en layout

[dry plover]

try localectl set-keymap de

[tired spire]

Done

[dry plover]

did that work?

[tired spire]

Yes

It created the subvolumes

[dry plover]

ls -al /mnt/nvme ?

[dry plover]

now unmount so we can mount the subvols sudo umount /mnt/nvme

[tired spire]

Done

[dry plover]

sudo mount -o subvol=@,noatime,compress=zstd /dev/nvme0n1p2 /mnt/nvme

sudo mkdir -p /mnt/nvme/{home,root,srv,var/cache,var/tmp,var/log,boot/efi}

[dry plover]

double check the syntax on this one, the commas, braces and stuff matter

[tired spire]

Done

[dry plover]

output of ls -al /mnt/nvme?

[dry plover]

do each of these individually

sudo mount -o subvol=@home,noatime,compress=zstd /dev/nvme0n1p2 /mnt/nvme/home
sudo mount -o subvol=@root,noatime,compress=zstd /dev/nvme0n1p2 /mnt/nvme/root
sudo mount -o subvol=@srv,noatime,compress=zstd /dev/nvme0n1p2 /mnt/nvme/srv
sudo mount -o subvol=@cache,noatime,compress=zstd /dev/nvme0n1p2 /mnt/nvme/var/cache
sudo mount -o subvol=@tmp,noatime,compress=zstd /dev/nvme0n1p2 /mnt/nvme/var/tmp
sudo mount -o subvol=@log,noatime,compress=zstd /dev/nvme0n1p2 /mnt/nvme/var/log
sudo mount /dev/nvme0n1p1 /mnt/nvme/boot/efi

[tired spire]

Done

[dry plover]

output of findmnt /mnt/nvme -R ?

[dry plover]

perfect

run each of these

sudo rsync -aHAX --info=progress2 /mnt/ssd/@/ /mnt/nvme/
sudo rsync -aHAX --info=progress2 /mnt/ssd/@home/ /mnt/nvme/home/
sudo rsync -aHAX --info=progress2 /mnt/ssd/@root/ /mnt/nvme/root/
sudo rsync -aHAX --info=progress2 /mnt/ssd/@srv/ /mnt/nvme/srv/
sudo rsync -aHAX --info=progress2 /mnt/ssd/@log/ /mnt/nvme/var/log/
sudo rsync -aHAX --info=progress2 /mnt/ssd/@tmp/ /mnt/nvme/var/tmp/

[tired spire]

With the different spacings?

[dry plover]

no

show the output of those after theyre done

[tired spire]

Yeah it takes a while
Goes from 98 to 97 somehow xD

And then to 94

99

97

lol

[dry plover]

lol

[tired spire]

Crazy progress tbh

Does the first command do the work of the others too?

[dry plover]

no, the first command is only stuff in the root subvol like bin, usr, etc directories

the rest of the commands fill in the rest

[tired spire]

damn didn't know how much stuff I have

[dry plover]

@home is probably going to be the longest, but the rest will be quick

[tired spire]

This is taking eternitys
Is there a way to after the first command ran continue tomorrow? My parents probably will tell me to sleep in about half an hour

[dry plover]

yes, its possible, but I'd avoid stopping in the middle of a command

[tired spire]

@dry plover I gotta go sleep now :(

[dry plover]

can you leave the computer on?

[tired spire]

It's kinda loud

[dry plover]

is that command finishing?

[tired spire]

Doesn't seem to finish

[dry plover]

press Ctrl C once

[tired spire]

Can a live iso hibernate?

[dry plover]

I don't think so

did Ctrl C stop it?

if so then do

sudo umount -R /mnt/nvme
sudo umount /mnt/ssd
sudo sync

[tired spire]

Ssd Target is busy

Oh I was in it

Now works

[tired spire]

Shutdown?

[dry plover]

not yet

oh did the umount work?

[tired spire]

Yes

[dry plover]

yes shutdown now

[tired spire]

Done

[dry plover]

tomorrow you can resume by remounting:

sudo mkdir -p /mnt/ssd /mnt/nvme
sudo mount -o ro,rescue=nologreplay,subvolid=5 /dev/sda2 /mnt/ssd
sudo mount -o subvol=@,noatime,compress=zstd /dev/nvme0n1p2 /mnt/nvme
sudo mkdir -p /mnt/nvme/{home,root,srv,var/cache,var/tmp,var/log,boot/efi}
sudo mount -o subvol=@home,noatime,compress=zstd /dev/nvme0n1p2 /mnt/nvme/home
sudo mount -o subvol=@root,noatime,compress=zstd /dev/nvme0n1p2 /mnt/nvme/root
sudo mount -o subvol=@srv,noatime,compress=zstd /dev/nvme0n1p2 /mnt/nvme/srv
sudo mount -o subvol=@cache,noatime,compress=zstd /dev/nvme0n1p2 /mnt/nvme/var/cache
sudo mount -o subvol=@tmp,noatime,compress=zstd /dev/nvme0n1p2 /mnt/nvme/var/tmp
sudo mount -o subvol=@log,noatime,compress=zstd /dev/nvme0n1p2 /mnt/nvme/var/log
sudo mount /dev/nvme0n1p1 /mnt/nvme/boot/efi

and then resuming the @ rsync and starting the rest of them:

sudo rsync -aHAX --info=progress2 /mnt/ssd/@/ /mnt/nvme/
sudo rsync -aHAX --info=progress2 /mnt/ssd/@home/ /mnt/nvme/home/
sudo rsync -aHAX --info=progress2 /mnt/ssd/@root/ /mnt/nvme/root/
sudo rsync -aHAX --info=progress2 /mnt/ssd/@srv/ /mnt/nvme/srv/
sudo rsync -aHAX --info=progress2 /mnt/ssd/@log/ /mnt/nvme/var/log/

[tired spire]

damn thank you so much

[static gust]

alpha.lynx received a thank you cookie!

[dry plover]

you're welcome!

feel free to ping me if needed

[tired spire]

without this discord I probably would have quit arch months ago, thanks

[dry plover]

oh and after finishing rsyncs (not pinging you so you can ignore till tomorrow), the next steps are to fix fstab:

sudo head -n 7 /mnt/nvme/etc/fstab > /tmp/h && sudo genfstab -U /mnt/nvme | cat /tmp/h - | sudo tee /mnt/nvme/etc/fstab

(this keeps the comment and regenerates the rest)
and make sure it's correct cat /mnt/nvme/etc/fstab should have entries for /, /home, /root, /srv, /var/cache, /var/tmp, /var/log, /boot/efi, all with the new btrfs UUID and the ESP with the new vfat UUID 4E39-D1E6.

then, check for the old UUID in any other config: sudo grep -rn '39f28f6e-510c-46f3-b2f9-4d351158485c\|F215-7CFF' /mnt/nvme/etc/ 2>/dev/null

then, chroot sudo arch-chroot /mnt/nvme and reinstall grub (I think you might've had grub?) grub-install --target=x86_64-efi --efi-directory=/boot/efi --bootloader-id=cachyos --recheck and grub-mkconfig -o /boot/grub/grub.cfg

then, regen initramfs mkinitcpio -P,

at this point it'd be good to sanity check bootloader exists, uefi boot entry exists with efibootmgr, bootloader config, initramfs, kernels etc. if all is well then should be good to unmount and reboot to nvme.

[tired spire]

rsync every command at the same time for more speed or after each other?

I'm at the first command and how is it already using that much my old SSD wasn't that big

[dry plover]

One rsync after the other, it won’t be any faster at the same time, they’re limited by the drive speed

[dry plover]

df doesn’t measure btrfs size correctly I think

[tired spire]

And rsync is only at 40%

It's at 400GB used I'm kinda worried

[tired spire]

btrfs should do it correctly I think

Command should be correct tho

[tired spire]

@dry plover that probably is not good

[dry plover]

hold on I just woke up

[tired spire]

Oh sorry

[dry plover]

no worries lol

[tired spire]

Oh I think 700GB .snapshots is the biggest enemy

[dry plover]

is it saying there's more usage than there was data on the ssd?

[tired spire]

On the nvme it has more usage than the whole SSD's size

[dry plover]

which rsyncs have finished so far?

[tired spire]

Damn

[tired spire]

First one I wasn't home

[dry plover]

run this sudo du -sh /mnt/ssd/@ /mnt/ssd/@home /mnt/ssd/@root /mnt/ssd/@srv /mnt/ssd/@cache /mnt/ssd/@tmp /mnt/ssd/@log

and then sudo du -sh --one-file-system /mnt/nvme /mnt/nvme/home /mnt/nvme/root /mnt/nvme/srv /mnt/nvme/var/cache /mnt/nvme/var/tmp /mnt/nvme/var/log

and sudo btrfs subvolume list /mnt/nvme

and show me the output of all of them

[dry plover]

oh just saw this

yeah snapshots utilize CoW so they don't take up much extra space, but rsync'ing bloats each snapshot

[tired spire]

So just continue rsyncing?

Because the command stopped at 71%

[dry plover]

lets delete the snapshots so we can reclaim space

the rsync stopping at 71% is expected because it was resumed, 29% was already transferred

[tired spire]

rm -r /mnt/nvme/.snapshots

??

[dry plover]

yes sudo rm -rf /mnt/nvme/.snapshots

[tired spire]

Damn my pc is transforming into a jet engine again

[dry plover]

also this makes me wonder if ssd originally had snapshots as a subvol. can you show sudo btrfs subvolume list /mnt/ssd | grep -i snap

[tired spire]

sudo btrfs subvolume delete path

Should I use that maybe?

Because RM can take ages

Ah it finished

Nice

[tired spire]

Finished

[dry plover]

ok now the nvme should have snapshots subvol sudo btrfs subvolume create /mnt/nvme/.snapshots

we might need to fix snapper config later

next check sudo btrfs filesystem usage /mnt/nvme

[tired spire]

28gigs is okay

[dry plover]

yes

proceed with the next rsync

[tired spire]

I'm finished with the mkinitcpio -P now

Not a single error

And 0 snapshots detected

unmount & reboot? @dry plover

[dry plover]

One sec

[tired spire]

Again, thank you so much for helping me

[static gust]

alpha.lynx received a thank you cookie!

[dry plover]

show output of efibootmgr -u

[dry plover]

I think we're good to unmount everything and reboot

[tired spire]

sudo umount -r /mnt/nvme
sudo umount -r /mnt/ssd

???

[dry plover]

captial R -R

[tired spire]

sudo umount -R /mnt/nvme
sudo umount -R /mnt/ssd

??

[dry plover]

yes

[tired spire]

Like that?

[dry plover]

perfect UEFI entry is there, yes boot that

[tired spire]

It boots

[dry plover]

awesome!

try snapper, we messed with its setup so it might error

[tired spire]

Thank you so much, without you I would have lost really much stuff?

[static gust]

alpha.lynx received a thank you cookie!

[dry plover]

you're welcome!

[dry plover]

check sudo btrfs subvolume show /.snapshots
and sudo snapper -c root list
and sudo snapper -c root list-configs

and I think snap-pac is already installed on cachyos so try pacman installing something to see if it snapshots

[dry plover]

perfect

[tired spire]

Nice, thanks man

[static gust]

alpha.lynx received a thank you cookie!