As discussed on ATR-26394, we needed to avoid writing every event processed by the beamspot chains using the full-scan tracking instance to the BeamSpotPEB stream, as this results in excessive bandwidth usage.
We address this by:
- applying a reject-only HypoTool to the instances that run the
beamspotVtx
configuration - adding a copy of each selection that does not run the beamspot algs, but does output its selected events to the BeamSpot stream.
A couple of minor details:
- I do find it slightly odd that we have a 'streamer' tool that rejects, but as the functionality already exists and it is arguably reasonable here because of the simplicity of the tool, it didn't make sense to reimplement. We could consider renaming, but for the vast majority of cases, it is being used as a streamer.
- We could consider renaming the chains to use the explicit roiftf selection rather than
j0_pf_ftf_presel(.*)
, but this naming scheme is easier to keep in sync with the primary chains defining the preselection choices.
Currently validating that the decisions are as intended.