Why Messages Containing FileNotFoundException Frequently Displayed in the HMaster Logs?

Question

Why messages containing FileNotFoundException and no lease are frequently displayed in the HMaster logs during the WAL splitting process?

2017-06-10 09:50:27,586 | ERROR | split-log-closeStream-2 | Couldn't close log at hdfs://hacluster/hbase/data/default/largeT1/2b48346d087275fe751fc049334fda93/recovered.edits/0000000000000000000.temp | org.apache.hadoop.hbase.wal.WALSplitter$LogRecoveredEditsOutputSink$2.call(WALSplitter.java:1330)
java.io.FileNotFoundException: No lease on /hbase/data/default/largeT1/2b48346d087275fe751fc049334fda93/recovered.edits/0000000000000000000.temp (inode 1092653): File does not exist. [Lease.  Holder: DFSClient_NONMAPREDUCE_1202985678_1, pendingcreates: 1936]
?at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.checkLease(FSNamesystem.java:3432)
?at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.analyzeFileState(FSNamesystem.java:3223)
?at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.getNewBlockTargets(FSNamesystem.java:3057)
?at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.getAdditionalBlock(FSNamesystem.java:3011)
?at org.apache.hadoop.hdfs.server.namenode.NameNodeRpcServer.addBlock(NameNodeRpcServer.java:842)
?at org.apache.hadoop.hdfs.protocolPB.ClientNamenodeProtocolServerSideTranslatorPB.addBlock(ClientNamenodeProtocolServerSideTranslatorPB.java:526)
?at org.apache.hadoop.hdfs.protocol.proto.ClientNamenodeProtocolProtos$ClientNamenodeProtocol$2.callBlockingMethod(ClientNamenodeProtocolProtos.java)
?at org.apache.hadoop.ipc.ProtobufRpcEngine$Server$ProtoBufRpcInvoker.call(ProtobufRpcEngine.java:616)
?at org.apache.hadoop.ipc.RPC$Server.call(RPC.java:973)
?at org.apache.hadoop.ipc.Server$Handler$1.run(Server.java:2260)
?at org.apache.hadoop.ipc.Server$Handler$1.run(Server.java:2256)
?at java.security.AccessController.doPrivileged(Native Method)
?at javax.security.auth.Subject.doAs(Subject.java:422)
?at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1769)
?at org.apache.hadoop.ipc.Server$Handler.run(Server.java:2254)

?at sun.reflect.GeneratedConstructorAccessor40.newInstance(Unknown Source)
?at sun.reflect.DelegatingConstructorAccessorImpl.newInstance(DelegatingConstructorAccessorImpl.java:45)
?at java.lang.reflect.Constructor.newInstance(Constructor.java:423)
?at org.apache.hadoop.ipc.RemoteException.instantiateException(RemoteException.java:106)
?at org.apache.hadoop.ipc.RemoteException.unwrapRemoteException(RemoteException.java:73)
?at org.apache.hadoop.hdfs.DataStreamer.locateFollowingBlock(DataStreamer.java:1842)
?at org.apache.hadoop.hdfs.DataStreamer.nextBlockOutputStream(DataStreamer.java:1639)
?at org.apache.hadoop.hdfs.DataStreamer.run(DataStreamer.java:665)

Answer

During the WAL splitting process, the WAL splitting timeout period is specified by the hbase.splitlog.manager.timeout parameter. If the WAL splitting process fails to complete within the timeout period, the task is submitted again. Multiple WAL splitting tasks may be submitted during a specified period. If the temp file is deleted when one WAL splitting task completes, other tasks cannot find the file and the FileNotFoudException exception is reported. To avoid the problem, perform the following modifications:

The default value of hbase.splitlog.manager.timeout is 600,000 ms. The cluster specification is that each RegionServer has 2,000 to 3,000 regions. When the cluster is normal (HBase is normal and HDFS does not have a large number of read and write operations), you are advised to adjust this parameter based on the cluster specifications. If the actual specifications (the actual average number of regions on each RegionServer) are greater than the default specifications (the default average number of regions on each RegionServer, that is, 2,000), the adjustment solution is (actual specifications/default specifications) x Default time.

Set the splitlog parameter in the hbase-site.xml file on the server. Table 1 describes the parameter.