Collect and handle the NameNode single node data block

We know that only one NameNode is active in the hdfs cluster, and only one NameNode provides external services at any time. Even when the HA mode is set up, one NameNode is in the Active state and the other is in the standby idle state, so for Hdfs, a single-point NameNode is always a weak point of performance.
1. The DataNode will report the new, deleted, and invalid data block information in the DataNode in the heartbeat with the NameNode (the report of these information is incrementalIn the form of reporting to the NameNode), the local memory of the NameNode will maintain an object blockMaps of the total data block, this object will maintain the information of all the data blocks in the entire cluster, in addition, the NameNode will also maintain an invalid block in the cluster., the list of damaged blocks, you can see the list of data blocks recorded by the NameNode about the entire cluster, the list of damaged data blocks and other information on the NameNode user interface.From here, we can see that the NameNode node will maintain all the information of the data blocks of the entire cluster, including those data blocks corresponding to those files, and then which Datanode the corresponding data blocks belong to, etc. It is conceivable that if there are more small files in the entire cluster, which means that under the same total amount of data, the number of data blocks will increase, and the objects such as the memory variable blockMaps that store the data blocks in the NameNode node will become larger and larger. In addition, the interaction between the Datanode and the NameNode will be required.The number of data blocks will also increase, which will lead to increased memory pressure in the NameNode and frequent gc, so this is why Hdfs does not want too many small file objects in the cluster.
In addition, hdfs fsckCommands can interact with the NameNode to output the disk block to which the file internally recorded in the NameNode belongs, the location of the damaged disk block, the DataNode to which a disk block belongs and the corresponding rack, etc., which greatly facilitates the administrator's operation and maintenance operations..