2026 Week 4
Checkout this week’s weeknote, with all the weekly happenings from week 4.
🆕 New Member
Warm welcome to our latest FDRI dev team member, Leanne Durkan! Joining us from Wallingford as a Research Software Engineer.
🌩️ AWS migration
After even more back and forth on how we structure our AWS accounts, we are back to having 2 AWS accounts, dri-staging and dri-production to support FDRI and other future relevant projects.
We have made the decision to do all of the migration in house, once the accounts are setup and we have access we can spin up our terraform and get started.
Current Plan (being actively worked on)
- New gitub repos for new accounts
- copy and paste existing terraform
- rename s3 buckets etc
- Run terraform locally to create infra
- Setup atlantis
- configure atlantis in github
- Update code pipelines to deploy to new accounts
🧑🤝🧑 Persona creation in progress
An effort to create user persona’s in underway, this will hopefully help guide our external comms on the projects as well as develop the requirements for our public api, user interface and code tools.
🐑 Custom authorizer setup
We are setting up a new url for basic authentication for MQTT to support some new IOT devices that don’t support certificate auth. It’s almost complete, see the diagram above for a little insight.
Update batch page and batch metadata
The batch page has been given a small makeover for a better user experience, allowing the last years worth of data to be plotted with the click of a button. Additionally, the page is now powered completely from the metadata store.
Flux pipeline
The flux/eddy covariance work is continuing to get rolling. This week we have spoken to Pete Levy, who a few years ago worked on eddystore which seems to have been aiming to solve some of the same problems we are looking at now. We’ve merged Ola’s ADR for the bucket layout, and Hamayoon has put together a proof-of-concept to build, configure and run EddyPro from Python code. We’re continuing to work out how to derive the processing configuration from the metadata service and how to integrate into the existing timeseries processor.