2023-05-18 bugfix release

[renatolfc]

Dear FarmVibes.AI Users,

We hope this message finds you well. We are thrilled to announce the latest release of FarmVibes.AI, which includes several exciting improvements, a new notebook that integrates with Azure Data Manager for Agriculture (ADMAg), and bug fixes that will enhance your experience with our platform.

In this release, we have fixed an important bug in our task orchestrator that affected fan-out/fan-in operations in workflows executed over the last month. This issue may have resulted in some discrepancies due to an ordering problem.

To maintain the accuracy and reliability of your data and workflows, we kindly suggest you upgrade to the latest version of FarmVibes.AI at your earliest convenience. As a precaution, we recommend rerunning workflows executed over the last month. The updated code will generate the correct outputs, only recalculating the problematic op outputs and their dependents.

As usual, if you already have cloned the repository:

git pull

should fetch the new code. We recommend an update of the library and images with:

bash farmvibes-ai.sh update

---

Here's a summary of the changes from the last pull request:

Workflows:

• [🎉 NEW] We added a new data_ingestion/admag/prescriptions workflow that fetches prescriptions from Azure Data Manager for Agriculture (ADMAg).
• [⚒️ FIX] We fixed an issue with wrong name of bands when running data_processing/index/index workflow on top of SpaceEye rasters.
• [⚒️ FIX] We fixed an issue where Sentinel-2 rasters were not having nodata values being set properly.

Notebooks:

• [🎉 NEW] We added a new notebook notebooks/heatmaps/nutrients-admag.ipynb that integrates the heatmap generation workflow with ADMAg.

Backend:

• [📈 IMPROVEMENT] We ported the backend of FarmVibes.AI to use asynchronous I/O instead of threads, improving performance and reducing memory usage.
• [📈 IMPROVEMENT] We improved the resiliency of our caching service, workers, and orchestrator. This will reduce the number of permanently stuck workflows due to transient errors, decrease memory leaks, and improve overall memory consumption.
• [⚒️ FIX] We fixed an issue when handling concurrent tasks in our fan-out/fan-in procedure, which caused subtasks results to be gathered in a random order. This would lead to incorrect results and caused tasks to be recomputed in every run.
• [⚒️ FIX] We fixed the order of initialization of containers in FarmVibes.AI, making new workflow run submissions work after cluster initialization/startup.