As an administrator you have sudo privileges on your research VM, which means you share the responsibility of keeping it healthy and secure with us. This page lists the things we ask you to do regularly, and the things we ask you not to do.

If you are unsure about anything on this page, please contact us before making the change.

Dos

Update your packages frequently

Keep the packages on your VM up to date, especially the security related ones. An unpatched VM puts your data, and the University network, at risk.

On Ubuntu VM, run:

sudo apt-get update
sudo apt-get upgrade

On Red Hat VM, run:

sudo yum update

Reboot the OS to apply the latest kernel

A newly installed kernel is only used after a reboot, so please reboot your VM after kernel updates. On Ubuntu, the file /var/run/reboot-required exists when a reboot is pending:

cat /var/run/reboot-required

Let the other users of the VM know beforehand, make sure no long running jobs are active, and then run:

sudo reboot

Notice: use reboot, not shutdown or poweroff. See Don’t shut off the VM below.

Clean up unused big files to free up space

A full disk can stop you from logging in and can break running jobs. Check the disk usage from time to time:

df -h

To find out what takes up the space in a directory, e.g. your home directory:

du -h --max-depth=1 ~ | sort -h

Typical candidates for a clean-up are old downloads and installers, intermediate results, unused conda environments and unused docker images. See also Disk or partition is full.

Put your important data on a volume or network drive rather than the root disk

We keep the root disk small on purpose. It holds the operating system, and it may be replaced when the VM is rebuilt or upgraded. Keep your research data on an external data volume or on a network drive such as your research drive, so that it is safe from system failures.

Back up your important data frequently

We don’t have automatic backup for research VMs. If a file is deleted by mistake, or a disk fails, we can’t restore it for you.

Copy your important data regularly to a place that is backed up, e.g. your research drive. For example, with a research drive mounted on /mnt/research_drive:

rsync -av /mnt/data/my_project/ /mnt/research_drive/my_project/

See also How to copy files from/to a Linux VM.

Don’ts

Don’t upgrade the operating system with do-release-upgrade

Please don’t move your VM to a new release of the operating system, e.g. by running do-release-upgrade on Ubuntu. Newer operating systems might not be supported by our management and security tooling yet, and a failed release upgrade can leave the VM unusable.

Regular package updates (see above) are fine and encouraged. If you need a newer release of the operating system, please contact us.

Don’t add users to, or remove users from, the VM

User accounts and access to the VM are managed by us. Please don’t run commands like adduser, useradd, deluser or userdel, and don’t edit the related system files by hand. If someone needs access to your VM, or should no longer have it, please contact us.

Don’t change the firewall rules to allow ingress

Please don’t change iptables, or other firewall rules (ufw, firewalld, nftables), to allow incoming connections. This might be insecure as it can expose your VM and your data. On top of that, the system automatically restores the original rules, so your change won’t last.

If your research needs a service on the VM to be reachable from other machines, please contact us.

Don’t install unnecessary software

Please don’t install software like an office suite or a web browser. A VM is not a replacement for your desktop computer. Every additional package takes up space on the small root disk, and needs to be kept up to date to stay secure.

Install only what your research needs, and remove software you no longer use.

Don’t shut off the VM

Please don’t shut off the VM, e.g. with sudo shutdown, sudo poweroff or sudo halt. You can’t turn it back on by yourself, and the VM stays unavailable for all its users until we start it again.

To restart the VM, use sudo reboot instead. If your VM has been shut off by mistake, please contact us.