[HDFS-13730] BlockReaderRemote.sendReadResult throws NPE

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 3.0.0
• Fix Version/s: 2.10.0, 3.0.4, 3.3.0, 2.8.6, 3.2.1, 2.9.3, 3.1.3
• Component/s: hdfs-client

Environment

Hadoop 3.0.0, HBase 2.0.0 + HBASE-20403.

(hbase-site.xml) hbase.rs.prefetchblocksonopen=true

• Assignee: Yuanbo Liu
• Reporter: Wei-Chiu Chuang
• Created: 11/Jul/18 20:11
• Updated: 02/Oct/19 17:15
• Resolved: 18/Jun/19 03:22

Description

Found the following exception thrown in a HBase RegionServer log (Hadoop 3.0.0 + HBase 2.0.0. The hbase prefetch bug HBASE-20403 was fixed on this cluster, but I am not sure if that's related at all):

2018-07-11 11:10:44,462 WARN org.apache.hadoop.hbase.io.hfile.HFileReaderImpl: Stream moved/closed or prefetch cancelled?path=hdfs://ns1/hbase/data/default/IntegrationTestBigLinkedList_20180711003954/449fa9bf5a7483295493258b5af50abc/meta/e9de0683f8a9413a94183c752bea0ca5, offset=216505135,
end=2309991906
java.lang.NullPointerException
at org.apache.hadoop.hdfs.net.NioInetPeer.getRemoteAddressString(NioInetPeer.java:99)
at org.apache.hadoop.hdfs.net.EncryptedPeer.getRemoteAddressString(EncryptedPeer.java:105)
at org.apache.hadoop.hdfs.client.impl.BlockReaderRemote.sendReadResult(BlockReaderRemote.java:330)
at org.apache.hadoop.hdfs.client.impl.BlockReaderRemote.readNextPacket(BlockReaderRemote.java:233)
at org.apache.hadoop.hdfs.client.impl.BlockReaderRemote.read(BlockReaderRemote.java:165)
at org.apache.hadoop.hdfs.DFSInputStream.actualGetFromOneDataNode(DFSInputStream.java:1050)
at org.apache.hadoop.hdfs.DFSInputStream.fetchBlockByteRange(DFSInputStream.java:992)
at org.apache.hadoop.hdfs.DFSInputStream.pread(DFSInputStream.java:1348)
at org.apache.hadoop.hdfs.DFSInputStream.read(DFSInputStream.java:1312)
at org.apache.hadoop.crypto.CryptoInputStream.read(CryptoInputStream.java:331)
at org.apache.hadoop.fs.FSDataInputStream.read(FSDataInputStream.java:92)
at org.apache.hadoop.hbase.io.hfile.HFileBlock.positionalReadWithExtra(HFileBlock.java:805)
at org.apache.hadoop.hbase.io.hfile.HFileBlock$FSReaderImpl.readAtOffset(HFileBlock.java:1565)
at org.apache.hadoop.hbase.io.hfile.HFileBlock$FSReaderImpl.readBlockDataInternal(HFileBlock.java:1769)
at org.apache.hadoop.hbase.io.hfile.HFileBlock$FSReaderImpl.readBlockData(HFileBlock.java:1594)
at org.apache.hadoop.hbase.io.hfile.HFileReaderImpl.readBlock(HFileReaderImpl.java:1488)
at org.apache.hadoop.hbase.io.hfile.HFileReaderImpl$1.run(HFileReaderImpl.java:278)
at java.util.concurrent.Executors$RunnableAdapter.call(Executors.java:511)
at java.util.concurrent.FutureTask.run(FutureTask.java:266)
at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.access$201(ScheduledThreadPoolExecutor.java:180)
at java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(ScheduledThreadPoolExecutor.java:293)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
at java.lang.Thread.run(Thread.java:748)

The relevant Hadoop code:

BlockReaderRemote#sendReadResult

void sendReadResult(Status statusCode) {
  assert !sentStatusCode : "already sent status code to " + peer;
  try {
    writeReadResult(peer.getOutputStream(), statusCode);
    sentStatusCode = true;
  } catch (IOException e) {
    // It's ok not to be able to send this. But something is probably wrong.
    LOG.info("Could not send read status (" + statusCode + ") to datanode " +
        peer.getRemoteAddressString() + ": " + e.getMessage());
  }
}

So the NPE was thrown within a exception handler. A possible explanation could be that the socket was closed so client couldn't write, and Socket#getRemoteSocketAddress() returns null when the socket is closed.

Suggest check for nullity and return an empty string in NioInetPeer.getRemoteAddressString.