Execute SQL times below error, baidu has the following several ways, the first: replicas too large (verified replicas less than the number of nodes); Second: the namenode datanode can not connect to normal (verified normal connection); Third: cluster space is not enough, the space has been used by 80%, so try to delete some data, found deleted after an error no longer, so that the problem settled, after more than a week and submitted to the same fault, has been 70% use space at this moment, and then I delete some data, found that is not an error again, after more than a week and at the same fault, has been 60% use space at this moment, the question becomes, 40% space can use is there, why would this mistake, and delete the part of data also can solve the problem, and every time an error, less and less space has been used,