Hi @doncucumber , Now you can have a good rest 🙂 Please check this article for the detailed HA concepts: https://docs.cloudera.com/documentation/enterprise/6/6.3/topics/cdh_hag_hdfs_ha_intro.html Active NN will sync up edit logs to majority of JournalNodes, so standby NN is capable of reading the edits from JNs. From your NN log we could see the reason of recovery failure is the recovery time exceeds the timeout 120000ms for a quorum of nodes to respond, so that's why I requested JN logs to check if edits sync-up failed and we found JN2 & 3's problem.   Regards, Will ... View more

Hi @Harsh J ,  I just deleted around 80% of my data with "DELETE from table_name where register <= '2018-12-31'" My disks are pretty full (around 90%). After the deletion nothing happened (about freeing space). I restart Cloudera (Kudu, Impala, HDFS, etc.) and nothing. I add this two lines to Kudu configuration (in "Master Advanced Configuration Snippet (Safety Valve) for gflagfile" and "Tablet Server Advanced Configuration Snippet (Safety Valve) for gflagfile"): ``` unlock_experimental_flags=true flush_threshold_secs=120 ``` After restart Kudu, wait for the 120 secs.. nothing. ... View more