30 s boot stall "Scanning for Btrfs filesystems"

[jlittle]

The NVMe SSD in my desktop died a few days ago, while it was in use (as the sender of a send/receive incremental backup). It was only a few years old. It is completely lifeless.

Using an old (2015) SATA SSD, I restored from a backup, and with some adjustments to /etc/fstab and a few subvolume creations (for .cache and snapper snapshots) Kubuntu 26.04 is running, albeit more slowly.

But during the boot, before plasma starts there is a step that stalls for about 30 s:

Scanning for Btrfs filesystems
registered /dev/sdc1
registered /dev/sdb6
registered /dev/sda2

(Those devices are those with btrfs on them, the last being the drive being booted.)

Can anyone explain this step? Removing the failed drive had no effect. Another install, a vanilla Kubuntu 26.04, on the same SSD does not have the problem.

[die.boer]

what does

sudo btrfs filesystem show

show ?
also

systemd-analyze blame

and

systemd-analyze critical-chain

?

[oshunluvr]

Try rebuilding your initramfs, if you haven't already.

Also wondering if your swap was on the failed drive.

[jlittle]

Thank you.

what does

sudo btrfs filesystem show

show ?

It shows the three btrfs that are present.

also

systemd-analyze blame

That doesn't show the 30 s delay. The top item is 3.738s NetworkManager-wait-online.service

and

systemd-analyze critical-chain

Again, does not show the 30 s delay. It would appear that the stall happens before systemd starts.

[jlittle]

Thank you.

Try rebuilding your initramfs, if you haven't already.

That fixed it (sudo update-initramfs -u). A step to be added when restoring from backup.

Also wondering if your swap was on the failed drive.

It was indeed, but commenting that out was part of patching up /etc/fstab.

[oshunluvr]

Great! I love the easy ones.