运行级联代码时，有人可以帮助我解决以下错误吗？
当使用CoGroup()连接两个数据流时，会发生此错误。如果将这两个数据流分别输出到接收器，则可以。
我正在使用Cascading 2.0和Hadoop 0.20.2-CDH3u5。
12/12/07 13:52:46 INFO flow.Flow: [my_tst] parallel execution is enabled: true
12/12/07 13:52:46 INFO flow.Flow: [my_tst] starting jobs: 2
12/12/07 13:52:46 INFO flow.Flow: [my_tst] allocating threads: 2
12/12/07 13:52:46 INFO flow.FlowStep: [my_tst] at least one sink does not exist
12/12/07 13:52:46 INFO flow.FlowStep: [my_tst] source modification date at: Thu Dec 06 10:15:32 EST 2012
12/12/07 13:52:46 INFO flow.FlowStep: [my_tst] starting step: (2/2)
12/12/07 13:52:47 INFO hdfs.DFSClient: Created HDFS_DELEGATION_TOKEN token 24769 for wagnerj
12/12/07 13:52:47 INFO security.TokenCache: Got dt for hdfs://nyxxx002.lab.dev:8020/user/wagnerj/.staging/job_201212070841_0008;…..
12/12/07 13:52:47 WARN snappy.LoadSnappy: Snappy native library is available
12/12/07 13:52:47 INFO util.NativeCodeLoader: Loaded the native-hadoop library
12/12/07 13:52:47 INFO snappy.LoadSnappy: Snappy native library loaded
12/12/07 13:52:47 INFO mapred.FileInputFormat: Total input paths to process : 1
12/12/07 13:52:47 INFO flow.FlowStep: [my_tst] submitted hadoop job: job_201212070841_0008
12/12/07 13:53:17 INFO flow.FlowStep: [my_tst] starting step: (1/2) cascd2
12/12/07 13:53:17 INFO mapred.FileInputFormat: Total input paths to process : 1
12/12/07 13:53:17 INFO mapred.FileInputFormat: Total input paths to process : 36
12/12/07 13:53:17 INFO mapred.JobClient: Cleaning up the staging area hdfs://nyxxx002.lab.dev:8020/user/wagnerj/.staging/job_201212070841_0009
12/12/07 13:53:17 ERROR security.UserGroupInformation: PriviledgedActionException as:wagnerj@LAB.DEV (auth:KERBEROS) cause:org.apache.hadoop.ipc.RemoteException: java.io.IOException: org.apache.hadoop.ipc.RemoteException: token (owner=wagnerj@LAB.DEV, renewer=mapred, realUser=, issueDate=1354906367141, maxDate=1355511167141, sequenceNumber=24769, masterKeyId=356) can't be found in cache
at org.apache.hadoop.mapred.JobTracker.submitJob(JobTracker.java:3469)
at org.apache.hadoop.mapred.JobTracker.submitJob(JobTracker.java:3437)
at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
最佳答案
不能不看源代码就知道。
关于hadoop - 在CDH3上运行级联代码时出错-在缓存中找不到 token ，我们在Stack Overflow上找到一个类似的问题： https://stackoverflow.com/questions/13771354/