On SODIN01 exceptionally connect to a single input buffer
Up until this change, HLT 1 was connecting on the two buffers when running on SODIN01 because there only one Allen instance runs.
This fixes the observed dropped meps on SODIN01 which has there since last year. It did not happen all the time so there must be some conccurrency "issue" that triggers it occasionally.
See also the thread at https://lblogbook.cern.ch/HLT/269
Edited by Rosen Matev