Matthew, Jonathan D., Bruno, Chris, Dominick, Nick, Kanika, Ryan, Satish, Nate, Michael G., Joseph, Adam, Gavin
- Some requests for new datasets were raised in the all analysis meeting on friday - will be discussed below.
- SAM for users Marc (Paterno) and Robert will start on this very soon.
- Production tests are fully automated and have summary graphs now
Jenkins build (Jonathan)¶
Kicked off by discussions of upgrading dedicated build machine, turns out there are a set of build services managed which utilise the "Jenkins" system. He gave this a go and it seems to be lower effort than originally envisaged. He hopes that all builds will transition some time this week. This migration will help future proof us, as when slf7 exists a machine will become available on Jenkins. Build times vary by machine tend to fall between our dedicated machine time and the gpvm node.
First analysis dataset round-up¶
- FD data - pre-shutdown CAFs appearing now, the blinded files will be announced while the unblinded files will stay hidden.
- FD cosmic MC - all done
- FD cosmic 14DB MC - ready for LEM handed off to Chris
Starting on Post-shutdown FD¶
A release was made for running reconstruction of the post-shutdown FD data in: FA14-10-28a. This release is identical to FA14-10-28 but with bad channels v6. Joseph is working with Dominick to run reconstruction on these. In parallel Jeny will get started on keep-up, Bruno validated the S15-01-06 and S15-01-12 releases, both are good, processing will use the more modern release.
MC generation for is waiting on the DQ group to produces the sub-runs list (if need be it could start without it). We were told that Bernali expects that she will get back to that this morning get the list soon. In order to propagate Ryan's new "simulated.cycle" parameter we need to add it to the known meta-data (Adam in fact did this during the meeting), we then also need it to be added in run_nova_sam.
1 week 14DB data in todays tag S14-01-12, all ND activity trigger - Paola is running this now.
Geo jittered reprocessing¶
A fix for LEM mixing is available in the new art this means that we should remix the LEM files of the geo-jittered dataset. We should also re-mix the baseline sample for comparisons. Matthew will email Gavin about this.
Calibration keep-up update (Paola)¶
This is up and running and the datasets have been sent around. MC jobs are not on at the moment.
- The jobs that failed are with grid experts. The current situation is that 2 additional sites that can be used have been identified.
- According to Ken Herner and Joe Boyd no sites should be using sl5, fermigrid is the highest sl5 user at only 2%.
Optimising production (Dominick, Matthew)¶
There was a question about how to handle the parentage of PID file if the reco file doesn’t exist yet. Dominick will try and add this at run time. The aim here is to test this methodology with the next set of ND rapid turn around.
Q. Is fhicl clever enough to handle complicated dual-use case fhicls? Chris to dig this up
Metadata type updates ¶
- The previous problem was with simulated subrun
- Think about keeping reco.keepup