Skip to content

TrfTestsART: Start using CA based merging and tighten the diff-root check in test_trf_q445_aodmerge.sh

Alaettin Serhan Mete requested to merge amete/athena:23.0-art-test-fix-aodmerge into 23.0

I noticed that the legacy merging has been failing in this ART test for a while now due to a trigger configuration issue. Since we adopted the CA-based merging in production already, I'm updating the test to use it.

I also tighten the diff-root check to require an exact match between the AODs before and after merging. This seems to fail in the most recent nightly, where we seem to be dropping some empty dynamic branches:

Py:diff-root      WARNING the following variables exist only in the old file !
Py:diff-root      WARNING  - [InDetForwardTrackParticlesAuxDyn.TRTdEdx]
Py:diff-root      WARNING  - [InDetForwardTrackParticlesAuxDyn.TRTdEdxUsedHits]
Py:diff-root      WARNING  - [InDetForwardTrackParticlesAuxDyn.eProbabilityNN]
Py:diff-root      WARNING  - [InDetForwardTrackParticlesAuxDyn.nBC_meas]
Py:diff-root      WARNING  - [InDetForwardTrackParticlesAuxDyn.trackLink]
Py:diff-root      WARNING  - [InDetForwardTrackParticlesAuxDyn.truthMatchProbability]
Py:diff-root      WARNING  - [InDetForwardTrackParticlesAuxDyn.truthOrigin]
Py:diff-root      WARNING  - [InDetForwardTrackParticlesAuxDyn.truthParticleLink]
Py:diff-root      WARNING  - [InDetForwardTrackParticlesAuxDyn.truthType]

This is similar to what we've noticed/briefly discussed in ATLASRECTS-7656. This needs to be followed-up but perhaps just setting TakeItemsFromInput=True and ForceRead=True on the stream side don't suffice in some cases.

cc: @gemmeren @jmaurer

Edited by Alaettin Serhan Mete

Merge request reports