XDC Replication discrepancies

We have observed occasional replication issue. Our set-up is two clusters with XDC replication enabled on both these clusters. However we have observed some discrepancies in few workflows not being replicated from primary to secondary cluster. To provide more information the missing workflow is in await state, strangely we have seen multiple workflows for same workflow type in same await state being replicated successfully.We have initiated a dummy signal on primary cluster for missing workflow and yet it hasn’t been replicated onto secondary cluster.

We also have been observing state-transition count differences between active and replication cluster for the same workflow. e.g.: statetransition count on active cluster is 11 and count on replication cluster is 10.

Our server version is 1.21.5

Could you please provide any pointers or reference for solving this replication issues?