Fix trackingOnly workflows for phase2_hcal sub-era #16170
Conversation
For phase2_hcal, offlinePrimaryVertices depends (via ak4CaloJetsForTrk and caloTowerForTrk) hbhereco, which is in hcalGlobalRecoSequence (by default the dependence is on hbheprereco). Simplest fix is to move hcalGlobalRecoSequence inside globalreco_tracking just before vertexreco.
|
A new Pull Request was created by @makortel (Matti Kortelainen) for CMSSW_8_1_X. It involves the following packages: Configuration/StandardSequences @cmsbuild, @franzoni, @davidlange6 can you please review it and eventually sign? Thanks. cms-bot commands are list here #13028 |
|
type bugfix |
|
is there hbheprereco in phase2_hcal? |
|
@cmsbuild please test |
|
The tests are being triggered in jenkins. |
|
@slava77 for sure JetCore has to be reviewed, especially with the new pixel pitch that entered in the latest geometry for PhaseII. Yet it would be important to have a working trackingOnly sequence working in any case. |
My impression is "no", but @kpedro88 (or somebody else from HCAL) should confirm.
Phase2 tracking doesn't have jetCore iteration (nor pixel cluster splitting). |
|
@slava77 right now phase2_hcal is using the generic HcalUpgradeDataFrame workflow from 62XSLHC, which only has one step for reco. At some point I would like to switch to using the Phase1 workflow, but I decided to wait until it was fully operational. |
|
actually, my comment about jetCore is not relevant, apparently. |
|
Not for the immediate PR, still it is for the future. PhaseII tracking at some point has to be aligned to whatever PhaseI will end up be.... Unless we want to start completely from scratch..... |
This PR suggests a fix to an exception in e.g. 20824.1 workflow RECO step (phase2 D3, "trackingOnly" workflow).
It appears that for
phase2_hcalsub-era,offlinePrimaryVerticesdepends (viaak4CaloJetsForTrkandcaloTowerForTrk)hbhereco, which is inhcalGlobalRecoSequence(by default the dependence is onhbheprereco). ButhcalGlobalRecoSequenceis not included inglobalreco_trackingthat is run in "trackingOnly" workflows (it is inglobalrecosequence right afterglobalreco_tracking).Simplest fix (proposed here) is to move
hcalGlobalRecoSequenceinsideglobalreco_trackingjust beforevertexreco. Limited matrix works, so I'd expect the change to not do any harm.Tested in CMSSW_8_1_X_2016-10-09-2300, no changes expected in monitored quantities. Workflow 20824.1 works again.
@rovere @VinInn @ebrondol @kpedro88