I stop and start the cluster over the weekend. Every time the cluster is started, namenode goes into safemode and i manually force it out of safemode using the command sudo -u hdfs hdfs dfsadmin
-safemode leave. I tried to leave it in safemode for more than 24 hours to see if it comes out of that eventually but it didn't. The under replicated blocks have grown to 20K+ in Ambari dashboard. Not sure if both the issues are related but i would like to understand the root cause of this. Any insights would be highly appreciated.