[SPARK-33182] Using Delegation Tokens to access HBase through Spark (Java)

• Type: Question
• Status: Resolved
• Priority: Major
• Resolution: Incomplete
• Affects Version/s: 2.3.0
• Component/s: Spark Core
• Labels: hbase kerberos

Environment

Spark 2.3, RHEL 7.5, HDP 3

• Assignee: Unassigned
• Reporter: Rishi S Balajis
• Created: 19/Oct/20 15:18
• Updated: 12/Dec/22 18:11
• Resolved: 20/Oct/20 04:41

Description

I have a requirement to access a kerberized HBase through delegation tokens instead of keytab. I have generated the token using the token util API and also loaded it back into the UserGroupInformation. However, hasKerberosCredentials returns a false. I am looking for the  the right way to use a saved delegation token to access HBase . The code that I have currently looks like shown below. I see the error :   

client.RpcRetryingCallerImpl: Call exception, tries=6, retries=36, started=4850 ms ago, cancelled=false, msg=Connection closed, details=row 'employee,,' on table 'hbase:meta' at region=hbase:meta,,1.1588230740, hostname=myhost.com,16020,1602758016346, seqNum=-1

 

UserGroupInformation ugi;
 ugi = UserGroupInformation.getLoginUser();
 Credentials creds = new Credentials ();
 creds = Credentials.readTokenStorageFile (new Path("file:///zen-volume-home/tokenFile"),conf);
 
 System.out.println("TOKEN *********" + creds.getToken(new org.apache.hadoop.io.Text("hbase")));
 ugi.addToken(creds.getToken(new org.apache.hadoop.io.Text("hbase")));
 ugi.addCredentials(creds);

/* I do see the token getting printed. However, I am looking for information on how to use this ugi which has the token added to it, to access data on HBase.
I have tried doing this :  */

SQLContext sqlC = ugi.getLoginUser().doAs((new PrivilegedExceptionAction<SQLContext>() {
		            public SQLContext run() throws Exception {
		     
		     sparkconf = new SparkConf().setAppName("Hbase With Spark");
		     jsc = new JavaSparkContext(sparkconf);
		     hbaseContext = new JavaHBaseContext(jsc, conf);
		 sqlContext = new org.apache.spark.sql.SQLContext(jsc);		
		        		
		 String sqlMapping  = "name STRING all:name" ;	
		 Dataset<Row> dfEmp = sqlContext.read().format("org.apache.hadoop.hbase.spark")
		        				.option("hbase.columns.mapping", sqlMapping)		        				.option("hbase.table", "employee").load();
		        		        
		        		dfEmp.registerTempTable("empdata");  
		        		dfEmp.show();
		        		return sqlContext;
		            }
		        }));