Closed jmunroe closed 4 months ago
@jmunroe @yuvipanda @AIDEA775 Is the Jupyterhub instance ready? If so, could you please grant my GitHub ID access to the hub?
Hi @ktyle -- as soon as the hub is deployed, we will let you know. This hub is @AIDEA775 next task to pick up.
Hi @ktyle and @jmunroe! I changed the logo to the blue version (https://projectpythia.org/_static/images/logos/pythia_logo-blue-rtext.svg) because the white version is not visible with the white background.
The staging and prod hubs are deployed! prod: https://projectpythia.2i2c.cloud/ staging: https://staging.projectpythia.2i2c.cloud/
But before you try to log in, I will temporarily need admin permission in the GitHub organization to do the last step:
Hi @AIDEA775 sounds good ... regarding the admin permission, is this something you need our organization (ProjectPythia) to provide, or are you waiting for someone in your 2i2c org to do this?
@ktyle yes, you'd have to grant that to him!
So close to complete :)
ok @AIDEA775 give it a try and let me know! Thanks!
Thank you @ktyle! Everything is ready. Can you log in?
@AIDEA775 nice!! Yes, I can log into both instances! Can any member of the ProjectPythia organization now log into these hubs, or do we need to provide a list of users for you?
Looks like folks can log in so I think we're good! Thanks again!
Thanks @AIDEA775 !!
We have Cloudbank funding to deploy a Project Pythia/Pangeo JupyterHub and BinderHub on GCP.
Cluster deployment:
This is flowing from a grant from NSF:
Rose, Brian E. J., Clyne, John, May, Ryan, Munroe, James, Snyder, Amelia, Eroglu, Orhan, & Tyle, Kevin. (2023). Collaborative Research: GEO OSE TRACK 2: Project Pythia and Pangeo: Building an inclusive geoscience community through accessible, reusable, and reproducible workflows. Zenodo. https://doi.org/10.5281/zenodo.8184298
This JupyterHub will support Project Pythia workflows for use by the broader Pangeo community.
I think we'll want both a persistent storage JupyterHub and ephemeral storage BinderHub both running on the same cluster. This is a pattern of JupyterHub for a smaller set of authorized, core users who create content, and a BinderHub for a much larger user group who do not require persistent storage.
We have a few projects with this access pattern requirement (HPOSS, HHMI, ProjectPythia, Catalyst) so it make some sense to first plan this kind of deployment as being just one instance of a JupyterHub/BinderHub combination on a dedicated cluster.
Phase 3 : Hub setup
There's going to be a number of hubs, and this starts specifying them.
Hub 1: Staging (Ready)
Phase 3.1: Initial setup
At the end of this phase, both 2i2c engineers and the admin users mentioned can login to the hub.
Phase 3.2: Object storage access
Phase 3.3: Profile List
These are the standard profile list options we want to start with. Can be customized later
unlisted_choice
deployer generate resource-allocation choices r5.xlarge --num-allocations 5
should be usedallowNamedServers
Phase 3.4: Authentication tuning (NOT READY)
Phase 3.5: Profile List finetuning
Hub 2: prod (Ready)
same as staging but for production
Hub 3: binder-like hub (Not Ready)
https://binder.projectpythia.2i2c.cloud
Equivalent functionality to https://binder.projectpythia.org
Authentication: requires a discussion with engineering on options.