Unfortunately, there's no simple way to resubmit all of the stuck tasks. In other cases where this has occurred, it often is counter productive to resubmit these tasks, as people have submitted new changes assuming that the original task is "dead" You can submit an idea for an enhancement request to have this considered for inclusion in a future version.
As for root cause, there are a couple of things to look at. First, what sort of tasks were running when this bottleneck occurred?
Was an explore and correlate just run? This is the most common cause of a large number of in progress tasks. When a E/C is run, every global user that is modified or created needs to callback to IM to get any updates back to the corporate user store. In this case a new IM task is created for every user that was affected by the E/C. Check the etanotify.log for any errors.
If you have not modified the JMS queue in JBoss to use a RDBMS, you should look at the tech doc that ashokpearl posted, as the JMS queue corruption in the default Hypersonic queues can cause problems with task data consistency, as all IM tasks are placed in the JMS queue for processing.
If this doesn't cover your situation, I would suggest that you open a support ticket and we can work further on the issue and get to resolution.
W.