Rescuing an unbootable Cloud Server

It’s pretty easy to mis-configure something and render one of your servers inaccessible or unbootable. Brightbox makes it easy to boot into a system rescue tool and fix the problem.

Graphical console

Firstly, open up the graphical console for your Cloud Server. You can do this with either Brightbox Manager or the command line interface.

Boot into the PXE system

PXE stands for “Preboot eXecution Environment”, and every Cloud Server can be made to boot into this instead of booting into the installed operating system. You just need to hit escape as it first starts to boot.

So hit the Send CtrlAltDel button in the graphical console to start a reboot of your system if you can. If you broke things so hard this doesn’t work, you’ll need to stop and start your server using the manager action menu (this kicks you out of the graphical console, so you’ll need to be quick to bring it back up in time after a start!)

Then hit the escape key when prompted, and you’ll be shown a selection of boot devices to choose from.

Virtio disk is your cloud server’s disk, which is the default to boot from normally. We want iPXE, so select 2.

The system will then boot into the iPXE system.

If you want to choose your own specific system rescue tool then hit Ctrl-B here. Otherwise, just wait and a default tool will boot for you.

The Brightbox default rescue tool is the system rescue cd, so if you’re familiar with that then you’re all set.

As it boots, it’ll ask which keyboard mapping you want. The default is US, but you can choose any of the available maps. Type uk and hit enter to get a United Kingdom keyboard map.

Then wait for the rescue image to be downloaded and booted. Once complete, you’ll see a shell prompt and you’re ready to do some rescuing:

Access your servers filesystem

Now that SystemRescue is booted, you can mount your Cloud Server’s filesystem and fix whatever you broke. The server’s disk is /dev/vda and most of our Linux images have the first partition, /dev/vda1, as the main filesystem.

So just make a mount point and mount it:

$ mkdir /mnt/vda1
$ mount /dev/vda1 /mnt/vda1

If you’re rescuing a Windows server, then it’s probably partition 2, so /dev/vda2.

If you have a more advanced storage layout like LVM then it’s a little trickier, but SystemRescue should have all the tools you need.

As an example, you could now change the password for a Linux user account with the chpasswd tool:

$ echo "ubuntu:anewpassword" | chpasswd -R /mnt/vda1

When you’re done, just type reboot and the filesystem will be unmounted and your server will reboot back into it’s operating system!

SSH access

SystemRescue starts up an SSH daemon by default, so just set a password by running passwd root and you can ssh into the rescue environment. Handy if you want to upload a new SSH key, or just want to just take a copy of your data from an unbootable system.