Rescue mode: repairing a server that won’t boot

Start the rescue system

In the client area, open your server, choose Rescue and confirm the reboot. The server starts a Debian-based live system in memory, reachable over SSH with your key (or the temporary password shown in the client area). Your disks are not touched until you mount them.

$ ssh root@203.0.113.10

Assemble and mount your disks

$ mdadm --assemble --scan          # bring up software RAID arrays
$ cat /proc/mdstat
$ lsblk -f                         # find the root filesystem
$ mount /dev/md2 /mnt              # adapt to your layout
$ for d in dev proc sys run; do mount --bind /$d /mnt/$d; done
$ chroot /mnt

Common fixes

Forgotten root password: passwd inside the chroot.
Broken boot loader (BIOS boot): grub-install /dev/nvme0n1 && grub-install /dev/nvme1n1 && update-grub.
Bad /etc/fstab entry: fix it, or comment the line out, then reboot.
Locked out by a firewall: disable the rule set, e.g. systemctl disable nftables.

Leave rescue mode

Exit the chroot, unmount everything, then disable rescue in the client area and reboot. The server starts from its own disks again.

$ exit
$ umount -R /mnt