Major Description
Right now when run with the -fix option, hbck tries to fix up the issue and then immediately re-runs itself to see if the fix worked. However most of the fixes require some other nodes in the cluster to take some action, which will take a couple of seconds (eg for them to notice a change in ZK and pick up the fixed region).
So, hbck should pause for some amount of time in between fixing and re-running.