docs

CANFAR Science Platform Documentation

Table of Contents

1. Introduction and Access
2. Interactive Sessions
3. Batch Jobs
4. Storage
5. Groups and Permissions
6. Programmatic Access
7. Community and Support
8. FAQ

Introduction and Access

Access to the CANFAR Science Platform is through an authorized access to the CANFAR Portal. A Canadian Astronomy Data Centre (CADC) Account is required.

• To request a CADC Account: https://www.cadc-ccda.hia-iha.nrc-cnrc.gc.ca/en/auth/request.html
• Authorization to access the science platform:
  • if you are part of a collaboration already using CANFAR, ask the admininistrator of the collaboration you belong to add you as a member of the group using the Group Interface
  • in any other case, send an email to [email protected] specifying you are requesting access to the CANFAR Science Platform and a short line of justification.

The CANFAR Science Platform runs software packaged in containers. The platform allows users to run both pre-built, shared containers or private, custom containers. Users can publish container images to the CANFAR Container Images Registry. We have specific documentation on how to build and publish capable to run on the CANFAR Science Platform.

The CANFAR Science Platform supports both launching interactive sessions and non-interactive ones. A more visual documentation for new users can be found here

Interactive Sessions

Interactive sessions are applications running server-side on the browser allowing users to interact with the (typically large) datasets hosted on CANFAR storage. There are a few types of Interactive Sessions on the platform:

Notebooks

Notebooks are using the Jupyter Lab interface.

CARTA

CARTA (Cube Analysis and Rendering Tool for Astronomy) is an astronomy visualization tool that will run natively in the browser. It can read FITS or HDF5 files, often used in radio astronomy, but not only.

Desktop

For running non browser-native applications in the Science Platform, a browser Desktop session can be launched.

• Desktop documentation and tutorials are described more in details in the User Documentation
• Launching a CASA window in the Desktop YouTube tutorial: YouTube Tutorial

Contributed

Contributed sessions are user-customised web applications, typically not maintained by CANFAR. This can be anything, such as a VSCode server and Pluto notebook for the Julia language.

The rules of building a container of type "contributed" on the CANFAR Science Platform are:

1. Incoming trafic will be over http (which may include websocket trafic) on port 5000
2. From the point of view of the container, requests will be received at the root path (/), but URLs in the browser will look like https:///, where and are subject to change. This path will initially be https://ws-uv.canfar.net/sessions/contrib/
3. The instance will be started by a script in the image that must be available at /skaha/startup.sh and will be passed 1 parameter: the sessionid.

Batch Jobs

There is limited possibility to run a batch job, which can be understood of a non-interactive executable launched on a headless container. Please contact us before making use of the headless job support--we are incrementally adding support for batch processing in the science platform. See the specific documentation. This is still experimental.

Storage

All sessions and applications accessed through Science Platform can share the same filesystem mounted at /arc. Within are /arc/home (contains all home directories) and /arc/projects (for project use). We encourage the use of /arc/projects for most data, and /arc/home for personalized configuration and software.

An efficient way to access the arc storage outside the Science Platform is through sshfs. Here is the documentation.

The arc storage is also accesible through an API. The following list the ways it can be accessed beyond the use of the Science Portal and sshfs:

• Using the CANFAR storage management interface: https://www.canfar.net/storage/arc/list
• Using the VOSPace Python libraries
• Using the /arc/files URL endpoint documentation

Please take care to protect sensitive information by ensuring it is not publicly accessible.

Groups and Permissions

Projects are encouraged to use groups to manage access to resources, including files and directories in arc project storage, mounted at /arc/projects in every session.

Groups and their memberships can be managed through the CANFAR groups web interface, here: https://www.cadc-ccda.hia-iha.nrc-cnrc.gc.ca/en/groups/

Once created, groups can be assigned to files and directories in arc storage directly from their interactive sessions, or through the CANFAR storage

For more details, see the documentation on file permissions

Programmatic Access

Session launching and management is through the skaha service. The skaha API definition and science platform service are here: https://ws-uv.canfar.net/skaha

Community and Support

Dicussions of issues and platform features take place in the Science Platform Slack Channel: Science Platform Slack Channel

Reporting of bugs and new feature requests:

• For the infrastructure and session https://github.com/opencadc/science-platform/issues
• For the containers: https://github.com/opencadc/science-containers/issues

Contributions to the platform (including updates or corrections to the documentation) can be submitted as pull requests to this GitHub repositories. We especially encourage science-containers.

General inquiries can be made to [email protected], and take a look at our FAQ.