Determine infrastructure needs for Pythia 2026 hackathon

[aprilmj]

Context

The Pythia hackathon will be in June for 2026 (earlier than previous years). We'll hear updates on any new / unique needs for this year's infrastructure after registration closes in late March.

Some infra notes from their 2025 post-mortem to help inform our follow up conversations:

• 2i2c infrastructure was appreciated, but was not stable (OSDF team)
• One had to re- start the server and use gh-scoped-creds to set up push access every time someone updated environment.yml (OSDF team)
• MyST issues, github tests passed even though some notebooks threw up errors (OSDF team)
• GitHub auth realistic solutions?

What we need to do

The most important thing to have done by May: know we can build cookbooks on the binder (complete the BinderBot implementation).

Based on the retro notes, it looks like there are also opportunities to improve the participant experience in advance.

Definition of Done

• The feature/service is technically complete
• The feature/service been tested with one or more users (if applicable)
• The feature/service been deployed to a production cluster
• The feature/service has been shown to be replicably deployable
• The feature/service follows our definition of minimally documented
• The feature/service has been added to our product menu (if applicable).
• Lessons learned have been documented
• The feature's availability has been communicated to Sales
• The feature has been communicated to our communities

[jmunroe]

From February The Project Pythia Cookoff 2026 planning meeting notes:

Current projectpythia.2i2c.cloud infrastructure:
RAM and CPU/GPU:

• 15.2, 30.3, 60.6,121.2 GB RAM, 15.6 CPU (r5-4xlarge)
• 16 GB RAM, 4 CPU, Tesla T4 GPU

2. Image Flavors:

• Pangeo Notebook image

• Jupyter SciPy Notebook

• Existing Docker container

• Build your own image (Binder)

• Pangeo PyTorch (GPU only)

• Pangeo TensorFlow

• Fri. 2/13: Kevin and Angus meet to go over clinder (updated cookbook actions, aka BinderbotNext)

• Fri. 3/6: Initial feedback on existing projectpythia.2i2c.cloud infrastructure

• What is the environment for each Pangeo-esque image flavor? Do they need updates?

• Do we want to only allow a Build-your-own image?

• To whom should we address questions regarding the projectpythia.2i2c.cloud infrastructure?

• Is binder.projectpythia.org an option we should advertise?

• James proposes on day 1 a standard catch-all environment as a base image. Groups can then add to their own custom image as their cookbooks take shape.

• James notes that we should have plenty of capacity on hand for the hackathon

• Binderbot-like functionality may not be needed at start of hackathon for most participants, but this is still something we want to have functioning well in advance of the hackathon, anticipating resource-intensive cookbooks (not specifically for the hackathon, it is for Project Pythia and needed now)

• Julia: what about environment caching (recall Tasha’s group from last year … very large environment that took a while to build … typically 30 minutes after each lunch) since the 2025 hackathon?

[jmunroe]

<Testing closing a CSH that does not have 'Hours spent' filled in>

[2i2c-Bot]

⚠️ This issue cannot be closed until Hours Spent is recorded

The Hours Spent field on the Product and Services project board has not been filled in.

Please update Hours Spent in the project board, then re-close this issue.

This issue was automatically reopened by csh-sync