Averaging it out on the last four hours vs the last four the instance was opprotunisticQM at the same time of day, processor use is actually up about 0.5% on average, but thats across stable processes not blocking IO, rather than the bloated spikes that inflate OQM's numbers.
There were an average of 1 session dropped on OQM as opposed to 6 on the daemons.
It seems the sessions get somehow affected by the queue? They shouldn't, but it seems the case. But ... correlation isn't causation. I'll have to try to get a ptrace on that bish