Bandwidth reduction for data taking after current/September MD

Thanks to @rjhunter for running some BW tests on mu=5.3 data. Unfortunately, we are about 1.8 GB/s over budget for the Turbo stream, so we need some reductions.

⚠ Rapid response is required since we plan start data taking at high mu around 1st October

However, applying slightly tighter pT/PID etc... cuts should be quite a safe and well understood change that can be done very quickly.

Here is are the top contributors to the Turbo BW:

Click to expand for the next 20 lines...

Full bandwidth pages can be found at https://rjhunter-bandwidth.web.cern.ch/HighMuData/

Log of the bandwidth test job: /afs/cern.ch/user/r/rjhunter/public/highmu_100k_2309.log

Moore snippet for local testing: https://gitlab.cern.ch/-/snippets/3269

Actions

• QEE WG Further prescale the QEE muon shower line
• Charm WG Try to tighten some cuts in the highest rate Turbo lines (see table below)
• B&Q WG Tighten the *EDM* lines
• RD WG Tighten Hlt2RD_BToHH_Incl ( !3859 (merged))
• BnoC Tighten Hlt2BnoC_Lb0ToPpPipPimPim_NoPID
• BnoC Remove Hlt2BnoC_BdToPipPim_NoPID from Turbo ( !3866 (merged))
• IFT Tighten Hlt2IFTTurbo_Femtoscopy_LcP

Impact on BW of changes thus far

(w.r.t. @rjhunter tests reported above)

Changes in units of MB/s.

WG	Merge request	Turbo (disk)	Full (tape)	Full (disk/spruce)
QEE	!3850 (merged)	-200	N/A	N/A
QEE	!3857 (merged)	N/A	approx. -900	approx. -20
IFT	!3867 (merged)	approx. -50	N/A	N/A
B&Q	!3883 (merged)	-	-	-
B&Q	!3854 (merged)	N/A	N/A	N/A
BnoC	!3866 (merged)	-20.	N/A	N/A
BnoC	!3856 (merged)	-20	N/A	N/A
Charm	!3869 (merged)	N/A	N/A	N/A
Charm	!3863 (merged)	N/A	N/A	N/A
Charm	!3871 (merged)	N/A	N/A	N/A
Charm	!3868 (merged)	N/A	N/A	N/A
Charm	!3862 (merged)	N/A	N/A	N/A
Charm	!3873 (merged)	N/A	N/A	N/A
SLB	!3882 (merged)	N/A	-2500	Unknown
SLB	!3879 (merged)	N/A	-200	Unknown

Table of highest BW charm/Turbo lines

Click to expand

Status as of 1st October RTA WP3 meeting

Early last week, with mu=5.3 data and 1.35 MHz Hlt1.

Click to expand

And as of yesterday with 1.2 MHz Hlt1.

Click to expand

• Thanks 👏 all for the fast response.
• Particular thanks to @rjhunter @tevans for running all BW the tests and HLT1 tunings, and of course @cagapopo for the pushing the release over the line.
• ⚠ some reductions on spruce/full lines may still be required by DPA

cc @decianm @mfontana @cagapopo @poluekt @msaur @lugrazet