All configured, the manual will be slaves can be the master start: 50070 and master: see 8088, but after all, one by one to start is very trouble, I do not know why not start together,

According to the information collected on the network, because namespaceID clusterID inconsistent, you need to change the same ID, this method tried, invalid,
And method is to delete all TMP logs to the format, the same GG,

View the log, doesn't even have a Warning
Time for a great god, and of all stripes talents!