POC AWS S3 pattern for Jupyter

[chrisbc]

AS NSHM team members

We want to share internal team data in amazon S3 buckets using

• [ ] google identity
• [ ] jupyter notebooks, hosted in mybinder.org

so that we can ensure our internal research data is available to all team members and not to the general public.

Example datasets:

• [ ] CFM 0.3 shapefiles
• [ ] CFM 0.3 tsurfaces
• [ ] CFM 0.3 crustal XML for opensha inversions

[chrisbc]

I've create a free tier amazon account for this POC. The aim is to use only free services, specifically:

• IAM (for user and access policies)
• S3 storage ( up to 5GB free/month)

REFS:

• https://pypi.org/project/s3contents/
• https://pypi.org/project/jupyters3/
• https://blog.codecentric.de/en/2018/04/accessing-aws-resources-with-google-sign-in/
• https://oscar6echo.gitlab.io/ipyauth/guide/google.html#overview ipyauth

[chrisbc]

@erabbott FYI this is the story we discussed earlier

[chrisbc]

Looks like this is not going to be possible with a free mybinder account.

Because the server URL is dynamic it will not be supported by the accounts.google.com OpenID service, which requires a stable URL for the RP application.

possible different options:

• [ ] obtains a stable jupyter notebooks domain (e.g nshm-nz.mybinder.org)
• [ ] create some sort of internediaty service that creates Amazon STS tiockets (or similar) to provide short-term access to the S3 resources. NB this would not be a great user experience.

[chrisbc]

ipyauth library is not working on local notebook setup. google API error: missing "response_type" parameter