Fix to run full HLT1 sequence when muons are in local

added rta-hotfix label

removed rta-hotfix label

added RTA label

added rta-operations label

awesome!

/ci-test

assigned to @rmatev

added ci-test-triggered label

[2023-05-16 23:16] Validation started with lhcb-master-mr#7885

@cagapopo do you have an idea why the Allen CI physics efficiency test fails? Could it be that these samples had events w/o muon raw banks, which had undefined behavior until now?

mentioned in merge request !1200 (merged)

I don't think this problem should be solved this way. Allen has a strong requirement that any input that is needed must exist and I think that should also hold for raw banks. If the muons are not in the partition, the sequence should not include the muons.

I understand the operational issues that result from this and they should be addressed, but in my opinion by configuring the correct sequence based on the partition content. Either fully through ECS or as part of the HLT1 production configuration by providing it with a dictionary of {partition_content: sequence, ..., default: default_sequence} and a way to know which detectors are in the partition.

If that is deemed too complex or can't be implemented for other reasons, skipping missing banks could be considered. At the very least, all algorithms that have outputs that depend on the banks being present should initialize their outputs to 0 (or another "valid when not available" value) when the banks are not there. MuonCalculateSRQSize doesn't currently do that, which leads to undefined behaviour. There could very well be more such issues. All algorithms that use muon hits have to be checked.