The TACH does not appear to be functioning as designed. I specify two separate machines (3 nodes each), but the backups are not distributed on the machines.
Here are TRACE logs (not of this system, unfortunately), but with the toString() of the CHF used. Here are the specific implementations:
session-cluster (session, accountToSessionId, clientToSessionId): TACHF
erm-cluster (serviceGroup, qamAllocation, eigAllocation): SyncTACHF