The namenode also knows the datanodes on which all the blocks for a given file are located, however, it does not store block locations persistently, since this information is reconstructed from datanodes when the system starts.
What does persistently means here ?
Here persistently means that name node would store information about block information in memory rather than maintaining those information in file system or HDFS or any data base or any kind of external storage.
Think of if it was persisted on file system, then the very next time when you boot your cluster you would have all the information. But do you think that the information that name node would have would be the latest one? Say a data node might be down due to hard ware failure hence the information with name node would be stale.