Issues with injected_finalVisitSummary on DRP-Prod pipeline

Support
#1

Hi, I’m using the d-latest DRP-Prod+injection pipeline and trying to run a personal subset of tasks that go like this

subset:
- makeDirectWarp
- selectGoodSeeingVisits
- templateGen
- getTemplate
- subtractImages
- makePsfMatchedWarp
- selectDeepCoaddVisits
- assembleDeepCoadd

This subset was originally made in the v24 of this pipeline, what I pasted above is in the same order as it was, with tasks renamed to match their current versions. When running it, the pipeline asks for injected_finalVisitSummary datasets, while I only have injected_visitSummary ones. Checking the pipeline, the only task that clearly outputs this type of dataset is makeDirectWarp, however even when ran by itself it still asks for injected_finalVisitSummary. Should I be adding some extra task to this subset with the new pipeline to get the datasets I need? Or can I just rename my injected_visitSummary with a connections command (I don’t quite understand the difference)?

Thank you for any help

makeDirectWarp issue
#2

Hmm, it’s a little hard to diagnose, but I’m guessing it has to do with changes in the names of datasetTypes produced by the pipelines between v24 and the latest pipelines. The unsatisfying answer would probably be to start from scratch so that all the processing takes place with the same pipelines version.

One thing you could try is visualizing the quantum graph by executing something similar to the commands at Getting started tutorial part 2: calibrating single frames with Pipeline Tasks — LSST Science Pipelines (to produce a version of Fig. 1 on that page, but for your pipeline). That might visually demonstrate where the injected_finalVisitSummary dataset is expected to be coming from.

#3

Hi @lawzul ! Just wanted to check in and see if Jeff’s suggestion solved your issue? If not, please let us know if there’s further information that might help us come up with a solution.

#4

I marked the above suggestion as the solution, since it seems fairly clear that the issue is related to mismatches between different pipelines versions and/or changes in data products between versions. However, do reach out if you are still unable to resolve your issues!