[HDFS-12383] Re-encryption updater should handle canceled tasks better

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 3.0.0-beta1
• Fix Version/s: 3.0.0-beta1
• Component/s: encryption
• Target Version/s: 3.0.0-beta1
• Hadoop Flags: Reviewed
• Assignee: Xiao Chen
• Reporter: Xiao Chen
• Created: 31/Aug/17 20:43
• Updated: 02/Oct/19 17:15
• Resolved: 04/Sep/17 02:51

Description

Seen an instance where the re-encryption updater exited due to an exception, and later tasks no longer executes. Logs below:

2017-08-31 09:54:08,104 INFO org.apache.hadoop.hdfs.server.namenode.EncryptionZoneManager: Zone /tmp/encryption-zone-3(16819) is submitted for re-encryption.
2017-08-31 09:54:08,104 INFO org.apache.hadoop.hdfs.server.namenode.ReencryptionHandler: Executing re-encrypt commands on zone 16819. Current zones:[zone:16787 state:Completed lastProcessed:null filesReencrypted:1 fileReencryptionFailures:0][zone:16813 state:Completed lastProcessed:null filesReencrypted:1 fileReencryptionFailures:0][zone:16819 state:Submitted lastProcessed:null filesReencrypted:0 fileReencryptionFailures:0]
2017-08-31 09:54:08,105 INFO org.apache.hadoop.hdfs.protocol.ReencryptionStatus: Zone 16819 starts re-encryption processing
2017-08-31 09:54:08,105 INFO org.apache.hadoop.hdfs.server.namenode.ReencryptionHandler: Re-encrypting zone /tmp/encryption-zone-3(id=16819)
2017-08-31 09:54:08,105 INFO org.apache.hadoop.hdfs.server.namenode.ReencryptionHandler: Submitted batch (start:/tmp/encryption-zone-3/data1, size:1) of zone 16819 to re-encrypt.
2017-08-31 09:54:08,105 INFO org.apache.hadoop.hdfs.server.namenode.ReencryptionHandler: Submission completed of zone 16819 for re-encryption.
2017-08-31 09:54:08,105 INFO org.apache.hadoop.hdfs.server.namenode.ReencryptionHandler: Processing batched re-encryption for zone 16819, batch size 1, start:/tmp/encryption-zone-3/data1
2017-08-31 09:54:08,979 INFO BlockStateChange: BLOCK* BlockManager: ask 172.26.1.71:20002 to delete [blk_1073742291_1467]
2017-08-31 09:54:18,295 INFO org.apache.hadoop.hdfs.server.namenode.ReencryptionUpdater: Cancelling 1 re-encryption tasks
2017-08-31 09:54:18,295 INFO org.apache.hadoop.hdfs.server.namenode.EncryptionZoneManager: Cancelled zone /tmp/encryption-zone-3(16819) for re-encryption.
2017-08-31 09:54:18,295 INFO org.apache.hadoop.hdfs.protocol.ReencryptionStatus: Zone 16819 completed re-encryption.
2017-08-31 09:54:18,296 INFO org.apache.hadoop.hdfs.server.namenode.ReencryptionHandler: Completed re-encrypting one batch of 1 edeks from KMS, time consumed: 10.19 s, start: /tmp/encryption-zone-3/data1.
2017-08-31 09:54:18,296 ERROR org.apache.hadoop.hdfs.server.namenode.ReencryptionUpdater: Re-encryption updater thread exiting.
java.util.concurrent.CancellationException
        at java.util.concurrent.FutureTask.report(FutureTask.java:121)
        at java.util.concurrent.FutureTask.get(FutureTask.java:192)
        at org.apache.hadoop.hdfs.server.namenode.ReencryptionUpdater.takeAndProcessTasks(ReencryptionUpdater.java:404)
        at org.apache.hadoop.hdfs.server.namenode.ReencryptionUpdater.run(ReencryptionUpdater.java:250)
        at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
        at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
        at java.lang.Thread.run(Thread.java:745)

Updater should be fixed to handle canceled tasks better.

Issue Links

• breaks: HDFS-12458 TestReencryptionWithKMS fails regularly
• is broken by: HDFS-10899 Add functionality to re-encrypt EDEKs