I am setting up HA singleton for our application so that we have a backup instance in the event of downtime.

I am testing a scenario where the current HA singleton provider (instance #1) is processing messages in the queue, and gets shut down. What happened was that I know there were  308 messages in the queue, and 70 got processed after the current singleton provider got shutdown. The new HA singleton provider (instance #2) does not process any messages until I started up instance #1. Then I can see the current HA singleton provider (instance #2) started to process messages, and the log file recorded 238 messages processed by instance #2

I was under impression that the messages are in the clustered queue, and should be available to any instances. However, this does not seem to be the case. Is there anyway to set it up so that the new HA singleton provider can continue process the rest of the messages?