Critical Time is being measured as the time between message send and message processing being finished.
This makes sense as a performance metric, since it includes queue wait time in the performance evaluation.
However, for delayed messages, the delay time is included into the critical time as well. And this is a bit strange to us. We like to monitor on critical time, because it provides a kind of SLA, but having the delay included in the critical time makes it useless for that purpose.
@jzeischka New releases of NServiceBus 7.7 and NServiceBus.Metrics 3.1 are now available that address this issue. Please review the closed issue for more details.