5-Minute Quick Start¶
Goal
By the end of this guide, you'll have a Jupyter Notebook Session on CANFAR with astronomy tools ready to use.
Prerequisites
- A CADC Account (Canadian Astronomy Data Centre) - Sign up here
- You have at least once logged into the CANFAR Science Platform and Harbor Container Registry.
- Python 3.10+
- Basic familiarity with Python and Jupyter notebooks
Installation¶
> pip install canfar --upgradeInstalled
Authentication¶
Login to CANFAR Science Platform
canfar auth login
canfar auth loginStarting Science Platform Login
Fetched CADC in 0.12s
Fetched SRCnet in 1.15s
Discovery completed in 3.32s (5/18 active)Select a Canfar Server: (Use arrow keys) 🟢 Canada SRCnet
🟢 UK-CAM SRCnet
🟢 Swiss SRCnet
🟢 Spain SRCnet
» 🟢 CANFAR CADCSelected a Canfar Server: 🟢 CANFAR CADCX509 Certificate AuthenticationUsername: usernameusername@ws.cadc-ccda.hia-iha.nrc-cnrc.gc.caPassword: ***********✓ Saving configuration
Login completed successfully!
Fetched CADC in 0.12s
Fetched SRCnet in 1.15s
Discovery completed in 3.32s (5/18 active)Select a Canfar Server: (Use arrow keys) 🟢 Canada SRCnet
🟢 UK-CAM SRCnet
🟢 Swiss SRCnet
🟢 Spain SRCnet
» 🟢 CANFAR CADCSelected a Canfar Server: 🟢 CANFAR CADCX509 Certificate AuthenticationUsername: usernameusername@ws.cadc-ccda.hia-iha.nrc-cnrc.gc.caPassword: ***********✓ Saving configuration
Login completed successfully!
Login Pathways
If you’re using the CADC CANFAR Science Platform and already have a valid certificate at ~/.ssl/cadcproxy.pem, the CLI will log in automatically
Starting Science Platform Login
✓ Credentials valid
✓ Authenticated with CADC-CANFAR @ https://ws-uv.canfar.net/skaha
Use --force to re-authenticate.
If you are a SRCnet user, you will be required to go through the OpenID Connect login process in your web browser.
Starting Science Platform Login
Fetched CADC in 0.13s
Fetched SRCnet in 1.03s
Discovery completed in 3.20s (13/19 active)
? Select a Canfar Server: 🟢 Canada SRCnet
Discovering capabilities for https://src.canfar.net/skaha
OIDC Authentication for https://src.canfar.net/skaha
Starting OIDC Device Authentication
✓ OIDC Configuration discovered successfully
✓ OIDC device registered successfully
✓ Follow the link below to authorize:
Force Re-Login
canfar auth login --force
What just happened?
canfardiscovered all available Science Platform servers around the world- You selected the
CADC CANFAR Server - You logged into the Science Platform using your CADC credentials
- The Science Platform generated a certificate for you valid for 30 days
- The certificate is stored in
~/.ssl/cadcproxy.pem
Launch Your First Notebook¶
Lets launch a Jupyter notebook with astronomy tools pre-installed,
# Launch a notebook sessioncanfar create notebook skaha/astroml:latestSuccessfully created session 'finish-inmate' (ID: d1tsqexh)
What just happened?
- We connected to CANFAR using your certificate
- The CLI defaulted the container image to
images.canfar.net/skaha/astroml:latest - A Jupyter notebook was launched with the container image in flexible mode
- A random name was generated for your session,
finish-inmatein this case - The Science Platform allocated flexible resources for your notebook and started it
Peek Under the Hood¶
# Timeline of events taken to launch the notebook sessioncanfar events $(canfar ps -q)
What just happened?
- We connected to CANFAR using your certificate
- We queried the Science Platform for all running sessions via
canfar ps -q - We fetched the events (actions performed by the Science Platform to start your session) for your session
- The events show the progress of your session being created
Check Status¶
canfar ps CANFAR Sessions
SESSION ID NAME KIND STATUS IMAGE CREATED
──────────────────────────────────────────────────────────────────────────────────────────
d1tsqexh finish-inmate notebook Running skaha/astroml:latest 7 minutes
SESSION ID NAME KIND STATUS IMAGE CREATED
──────────────────────────────────────────────────────────────────────────────────────────
d1tsqexh finish-inmate notebook Running skaha/astroml:latest 7 minutes
What just happened?
- We connected to CANFAR using your certificate
- The status of your session was checked
- The session is in
Runningstate, ready to use
Get Session Information¶
canfar info $(canfar ps -q)
Session ID d1tsqexh
Name finish-inmate
Status Running
Type notebook
Image images.canfar.net/skaha/astroml:latest
User ID brars
Start Time 13 minutes ago
Expiry Time 3 days and 23.77 hours
Connect URL https://connect.to/notebook/here
UID 123456789
GID 123456789
Groups [12345, 67890]
App ID <none>
CPU Usage 0% of 1 core(s)
RAM Usage 0% of 2G GB
GPU Usage Not Requested
Session ID d1tsqexh
Name finish-inmate
Status Running
Type notebook
Image images.canfar.net/skaha/astroml:latest
User ID brars
Start Time 13 minutes ago
Expiry Time 3 days and 23.77 hours
Connect URL https://connect.to/notebook/here
UID 123456789
GID 123456789
Groups [12345, 67890]
App ID <none>
CPU Usage 0% of 1 core(s)
RAM Usage 0% of 2G GB
GPU Usage Not Requested
What just happened?
- We connected to CANFAR using your certificate
- The information for your session was fetched
- When we created your session, we never specified a name, CPU or memory, so flexible mode was used
- Flexible mode allows your session to adapt its resource usage based on cluster availability
- The session lifetime defaults to 4 days
Access Your Notebook¶
Check the status and get the URL to access your notebook:
canfar open $(canfar ps -q)Opening session tcgle3m3 in a new tab.
What just happened?
- We connected to CANFAR using your certificate
canfar ps -qreturns only the session ID of your session- Your browser opened the notebook in a new tab
Pro Tip
The notebook usually takes 60-120 seconds to start. You can also check status from the command line:
Resource Allocation Modes¶
CANFAR Science Platform supports two resource allocation modes, see platform concepts for more information.
Start Analyzing!¶
Once your notebook is running, click the URL to open it in your browser. You'll have access to:
- Jupyter Lab with a full Python environment
- Pre-installed astronomy libraries: AstroPy, Matplotlib, SciPy, PyTorch, etc.
- Storages
- Persistent: Your work is automatically saved at
/arc/home/username/ - Project: Large datasets shared within your project at
/arc/projects/name - Ephemeral: For temporary data staging, use
/scratch/
- Persistent: Your work is automatically saved at
Try This First
In JupyterLab, open a new Notebook and run the following code to verify your environment:
import astropy
from astropy.io import fits
import matplotlib
import numpy as np
print(f"AstroPy version: {astropy.__version__}")
print(f"Matplotlib version: {matplotlib.__version__}")
print(f"Numpy version: {np.__version__}")
print(f"GPU available: {torch.cuda.is_available()}")
print("Ready for astronomy!")
Clean Up¶
When you're done, clean up your session to free up resources for others:
canfar delete $(canfar ps -q)Confirm deletion of 1 session(s)? [y/n] (n): y
Successfully deleted {'tcgle3m3': True} session(s).
Successfully deleted {'tcgle3m3': True} session(s).
Congratulations!¶
You now have a fully-equipped astronomy computing environment running in the cloud. No software installation, no environment conflicts, no waiting for local resources.
Troubleshooting¶
Common Issues
- Notebook won't start?
- Check available resources:
canfar stats - Try flexible mode (default) for faster scheduling
- If using fixed mode, try smaller resource values (fewer cores/RAM)
- Check session status:
canfar ps
- Check available resources:
- Can't access notebook URL?
- Wait 1-2 minutes for full startup
- Check if you're on a VPN that might block the connection
- Verify the session is in "Running" status
- Variable performance in flexible mode?
- This is normal - performance adapts to cluster load
- For consistent performance, use fixed mode with specific
--cpuand--memoryvalues