Errors retrieving items

We had 3 servers in a cluster. One went down and after it was brought back up we did not re-balance. Then within 24 hours a 2nd went down and was brought back up and we then initiated a re-balance. We are seeing intermittent items that appear to not be in the cache. On the cluster view, the item size is very high on the one that was up, but the other 2 are still very low. Should we re-balance, flush? What is the recommended next step?