There's a case where sending non tx messages, persistently, you may get out of sync on the counters with paging in case of a failure I never consider that a big deal since it would be a small window where the counter would be out of sync... after a failover until the page was completely consumed.. however users are complaining about this. I have implemented a new record type on page that will be a point of sync when a new counter is on a pending state regarding a page file. When the page is closed I then remove the pending record and make the cursor effective.
Verified with EAP 6.2.3.CP.CR2 / HornetQ 2.3.14.1.Final