I have an alert in Ambari saying there is missing 50 blocs since 5 august.
Using some existing post I found the command to identify the missing blocks: "sudo -u hdfs hdfs fsck /".
I have 2 questions:
1) The blocks missing does it mean that these blocks are missing in one node or in all nodes?
2) Some of these blocks are from HBase, so the solution to delete them cannot work for me. In the other hand in the HDFS log I can see synchronization for these Hbase missing blocks based on 1 august. Is there any way to restore these blocks?
1) the "blocks missing" it means that those blocks are not available in the live data nodes of the cluster, please verify any data node is down/unhealthy and bring it back - might rectify the missing blocks.
2) at hbase side if you had region replication >1 then there is a possibility of blocks availability in replicated region, else not possible.
the blow link might be helpful
Thank you for your response.
Something that i don't undrestand. Our replication factory is the default value, 3.
So how we can loss the same blocks on 3 nodes?
We have a replication factory of 2 forHBase. Let me go more deep on this part.