Seven steps from nothing to a running experiment. Each step links the one page that covers it in depth — this page is the map, not the territory.
Register at www.cosmos-lab.org/portal/register. You join (or create) an organization — this is your project group: it determines who approves your account, whose reservations you can share, and which resources you may use.
Approval is two-stage: first activate via the email link you receive, then wait for approval by your organization's PI or a testbed administrator. Whitelist support@orbit-lab.org (account emails) and scheduler@cosmos-lab.org (reservation emails) so nothing lands in spam.
→ Details: Portal: Getting Started
Generate a key if you don't have one (ssh-keygen -t ed25519), and upload the public key on your portal profile. All testbed access is key-based — there are no passwords on consoles.
→ Details: Portal: Account Management
Open the Scheduler and click a free future slot on a domain. For a first session pick a sandbox (small, quick to get) rather than the main grid. Reservations are auto-approved: your first slot of the day is confirmed by the nightly cycle, and anything else is approved just-in-time about 90 seconds before it starts. Maximum duration is per-domain (2–6 hours).
→ Details: Portal: Scheduler
When your slot starts, log into the reserved domain's console with your portal username:
ssh <username>@console.<domain>.cosmos-lab.org
The console is the jump host and control point for everything in the domain. Your home directory is shared across all consoles.
→ Details: SSH Access
From the console, load the baseline image onto a node and turn it on. omf load leaves nodes powered off when it finishes — powering on is a separate step:
omf load -t node1-1 -i baseline.ndz
omf tell on -t node1-1
Give it a minute to boot, then confirm you are on your image, not the PXE stage OS:
ssh root@node1-1 cat /.omf-image-name # should print baseline.ndz
For SDR work use -i baseline-sdr.ndz (UHD + GNU Radio preinstalled).
→ Details: OMF Quick Start · Disk Images
Nodes are yours as root (ssh root@node1-1). The fastest way to something real is the first Ansible experiment — one command that stages, runs, and collects a complete experiment. Every tutorial can be run three ways: manual CLI (see each command), Ansible (one reproducible run), or Jupyter (interactive notebook).
Copy anything you want to keep off the nodes and into your (shared, persistent) console home directory — node disks are wiped at the next imaging. Then be a good citizen:
scp root@node1-1:/root/results.csv ~/
omf tell offs -t all
If you customized a node and want to keep the setup, save it as your own disk image before the reservation ends.