Viraj Jasani created HDFS-16907:
-----------------------------------
Summary: BP service actor LastHeartbeat is not sufficient to track
realtime connection breaks
Key: HDFS-16907
URL: https://issues.apache.org/jira/browse/HDFS-16907
Project: Hadoop HDFS
Issue Type: Improvement
Reporter: Viraj Jasani
Assignee: Viraj Jasani
Attachments: Screenshot 2023-02-03 at 6.12.24 PM.png
Each BP service actor thread maintains _lastHeartbeatTime_ with the namenode
that it is connected to. However, this is updated even if the connection to the
namenode is broken.
Suppose, the actor thread keeps heartbeating to namenode and suddenly the
socket connection is broken. When this happens, until specific time duration,
the actor thread consistently keeps updating _lastHeartbeatTime_ before even
initiating heartbeat connection with namenode. If connection cannot be
established even after RPC retries are exhausted, then IOException is thrown.
This means that heartbeat response has not been received from the namenode. In
the loop, the actor thread keeps trying connecting for heartbeat and the last
heartbeat stays close to 1/2s even though in reality there is no response being
received from namenode.
Sample Exception from the BP service actor thread, during which LastHeartbeat
stays very low:
{code:java}
2023-02-03 22:34:55,725 WARN [xyz:9000] datanode.DataNode - IOException in
offerService
java.io.EOFException: End of File Exception between local host is: "dn-0";
destination host is: "nn-1":9000; : java.io.EOFException; For more details see:
http://wiki.apache.org/hadoop/EOFException
at sun.reflect.GeneratedConstructorAccessor34.newInstance(Unknown Source)
at
sun.reflect.DelegatingConstructorAccessorImpl.newInstance(DelegatingConstructorAccessorImpl.java:45)
at java.lang.reflect.Constructor.newInstance(Constructor.java:423)
at org.apache.hadoop.net.NetUtils.wrapWithMessage(NetUtils.java:913)
at org.apache.hadoop.net.NetUtils.wrapException(NetUtils.java:862)
at org.apache.hadoop.ipc.Client.getRpcResponse(Client.java:1553)
at org.apache.hadoop.ipc.Client.call(Client.java:1495)
at org.apache.hadoop.ipc.Client.call(Client.java:1392)
at
org.apache.hadoop.ipc.ProtobufRpcEngine2$Invoker.invoke(ProtobufRpcEngine2.java:242)
at
org.apache.hadoop.ipc.ProtobufRpcEngine2$Invoker.invoke(ProtobufRpcEngine2.java:129)
at com.sun.proxy.$Proxy17.sendHeartbeat(Unknown Source)
at
org.apache.hadoop.hdfs.protocolPB.DatanodeProtocolClientSideTranslatorPB.sendHeartbeat(DatanodeProtocolClientSideTranslatorPB.java:168)
at
org.apache.hadoop.hdfs.server.datanode.BPServiceActor.sendHeartBeat(BPServiceActor.java:544)
at
org.apache.hadoop.hdfs.server.datanode.BPServiceActor.offerService(BPServiceActor.java:682)
at
org.apache.hadoop.hdfs.server.datanode.BPServiceActor.run(BPServiceActor.java:890)
at java.lang.Thread.run(Thread.java:750)
Caused by: java.io.EOFException
at java.io.DataInputStream.readInt(DataInputStream.java:392)
at org.apache.hadoop.ipc.Client$IpcStreams.readResponse(Client.java:1884)
at
org.apache.hadoop.ipc.Client$Connection.receiveRpcResponse(Client.java:1176)
at org.apache.hadoop.ipc.Client$Connection.run(Client.java:1074) {code}
Attaching screenshots of how last heartbeat value looks when the above error is
consistently getting logged.
Last heartbeat response time is important to initiate any auto-recovery from
datanode. Hence, we should introduce LastHeartbeatResponseTime that only gets
updated if the BP service actor thread was successfully able to retrieve
response from namenode.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-dev-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-dev-h...@hadoop.apache.org
