The following script determines and returns whether there are events stuck -- meaning they were claimed by a node which, for whatever reason, is unavailable at the moment -- thus, event won't be processed.

How to use:

1. Execute as a background script / fix script

(function () {
   var gaEvent = new GlideAggregate('sysevent');
   gaEvent.addAggregate('COUNT', 'claimed_by');
   gaEvent.addQuery('state', '!=', 'processed');
   gaEvent.addQuery('state', '!=', 'error');
   gaEvent.addQuery('state', '!=', 'transferred');
   gaEvent.addNotNullQuery('claimed_by');
   gaEvent.query();

   while (gaEvent.next()) {
      var node = gaEvent.getValue('claimed_by');
      var count = gaEvent.getAggregate('COUNT', 'claimed_by');

      var grNode = new GlideRecord('sys_cluster_state');

      grNode.addQuery('status', 'online');
      grNode.addQuery('system_id', node);
      grNode.query();

      grNode.hasNext() ? 
      gs.log('No stuck events found') :
      gs.log('Unavailable node: ' + node + ' is claiming ' + count + ' events', 'stuck_event');
   }
})();

Disclaimer: Please, always test the script in non-productive environments before running in a productive instance. Script was reviewed at the creation time but may break/get unexpected results due to instance versioning and SN API upgrades.