[GitHub] [hbase] anoopsjohn commented on a change in pull request #2237: HBASE-24833: Bootstrap should not delete the META table directory if …

[GitBox <gi...@apache.org>]

anoopsjohn commented on a change in pull request #2237:
URL: https://github.com/apache/hbase/pull/2237#discussion_r469412600



##########
File path: hbase-server/src/main/java/org/apache/hadoop/hbase/master/HMaster.java
##########
@@ -915,6 +917,11 @@ private void finishActiveMasterInitialization(MonitoredTask status)
       this.tableDescriptors.getAll();
     }
 
+    // check cluster Id stored in ZNode before, and use it to indicate if a cluster has been
+    // restarted with an existing Zookeeper quorum.
+    isClusterRestartWithExistingZNodes =

Review comment:
       Ya at proc level..  But before that itself one more thing.
   Say there is a cluster recreate over existing data and so no zk node for clusterId. We will get true for 'isClusterRestartWithExistingZNodes '. In next lines, we will create and write the zk node for clusterId.  Now assume after executing that lines, the HM restarted.  So the zk node was created but the InitMetaProc was NOT submitted.  Now after restart, when we come here, we have zk data for clusterId.  So 'isClusterRestartWithExistingZNodes ' will become false.  Now this time the InitMetaProc started and as part of that we will end up deleting the Meta dir.
   So this says the need to keep this boolean info somewhere once we find that and even before creating the zk node for ClusterId.  Am I making the concern clear this time?




----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
users@infra.apache.org