I have an HDFS Cluster where I have stored a file that has a replication factor of 3. If one DataNode is down, I should still be able to read, delete and upload that file again. HDFS should recognize the failure of the DataNode, mark it as under-replicated and should start to create a replica on another DataNode.
What happens if two DataNodes fail at the same time and 2 out of 3 replicas are offline. Am I still able to read, delete and upload that file on it again? HDFS would not be able to write the changes on a majority of DataNodes but it would still be able to write the changes to the NameNode.