Closed Sohambutala closed 1 week ago
Attention: Patch coverage is 7.19719%
with 1586 lines
in your changes missing coverage. Please review.
Project coverage is 31.68%. Comparing base (
f85b12e
) to head (83bb7d1
). Report is 2 commits behind head on main.
:exclamation: There is a different number of reports uploaded between BASE (f85b12e) and HEAD (83bb7d1). Click for more details.
HEAD has 1 upload less than BASE
| Flag | BASE (f85b12e) | HEAD (83bb7d1) | |------|------|------| ||2|1|
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
Converting this PR to draft since Prefect variables have a limit of 255 characters. An alternative would be to try using the Echodataflow block object.
This PR has some files that seems from an older version. I point to one example in the comment above, but I also notice some of the actions versions got reversed to the older ones. So maybe you check this, before merging.
Thank you for highlighting workflow changes, I had those locally but never committed, since I used a different branch to merge into main after the hack with Caesar and Brandyn.
These extensions are designed to trigger or support the main pipeline, aiding in data source acquisition, enabling concurrent deployments etc.
1. File Monitoring for Automated Deployments:
Purpose: To monitor file changes and trigger specific deployments.
Current Approach:
Schedule: watchdog is not used currently as we only need to schedule at regular intervals. This can be changed later, or a separate extension can be added for real-time scenarios.
Deployment Independence: Each file triggers a new deployment, and a failure in one deployment does not affect others.
Implementation Details:
Tracking: Utilizes
Prefect Blocks
to record the last successful run time and the list of processed files.Concurrency: Leverages task concurrency to run multiple deployments simultaneously, allowing for concurrent file processing.
Fail-Safe Option: The
fail_safe
option toggles between concurrent and sequential processing. In sequential mode, a failure in one file stops the processing of remaining files. In concurrent mode, file processing is independent.Considerations: Concurrency must be managed carefully when appending results to a single Zarr store to avoid race conditions and maintain the order of data.
2. File Downloader for Seamless Data Transfer:
Purpose: To facilitate data transfer between two locations.
Implementation Details:
edf-data-transfer
).Delete Option: Offers an option to delete files after transfer, with a default setting of
false
.Considerations: Be mindful of rate limitations of both the source and destination. Concurrent requests can trigger rate limits.