...
File versioning in the L1 pipeline: The reason for this is that a run will usually be processed piecemeal (run cut by a downlink/data transfer, missing data showing up, etc). While we only process a any contiguous chunk of data once, we always merge up everything every time we process a piece of a run. It's the latest version of the merged file that will be the most complete and which will be of (most) interest. The infrastructure (Data catalog etc) needs to take this into account.
...
We have processed both MC and real data on the DEV/TEST server with this version of L1Proc.
Rollback procedure
Rolling Both L1Proc and infrastructure code can be rolled back to the previous version of L1Proc must be coordinated with similar rollback of infrastructure codeversions.
CCB Jira
Details
...