[
https://issues.apache.org/jira/browse/HBASE-17501?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
James Moore updated HBASE-17501:
--------------------------------
Attachment: HBASE_17501.patch
I've created an HFileUtil class to handle the DRY aspects. I'm somewhat
neutral on whether we should reseek on an IOE and added it for defensive
purposes rather than a particular need as different FileSystems could have
different semantics around seek and seekToNewSource.
The underlying implementation of seek on DFSIS appears only to release an IOE
when the stream is closed and swallows any other IOEs. If this is the desired
behavior of FSInputStream, we should be all set with just catching the
NullPointer exception.
updated patch file using HFileUtil while only catching NullPointers attached.
> NullPointerException after Datanodes Decommissioned and Terminated
> ------------------------------------------------------------------
>
> Key: HBASE-17501
> URL: https://issues.apache.org/jira/browse/HBASE-17501
> Project: HBase
> Issue Type: Bug
> Affects Versions: 1.2.0
> Environment: CentOS Derivative with a derivative of the 3.18.43
> kernel. HBase on CDH5.9.0 with some patches. HDFS CDH 5.9.0 with no patches.
> Reporter: Patrick Dignan
> Priority: Minor
> Attachments: HBASE_17501.patch, HBASE_17501.patch
>
>
> We recently encountered an interesting NullPointerException in HDFS that
> bubbles up to HBase, and is resolved be restarting the regionserver. The
> issue was exhibited while we were replacing a set of nodes in one of our
> clusters with a new set. We did the following:
> 1. Turn off the HBase balancer
> 2. Gracefully move the regions off the nodes we’re shutting off using a tool
> we wrote to do so
> 3. Decommission the datanodes using the HDFS exclude hosts file and hdfs
> dfsadmin -refreshNodes
> 4. Wait for the datanodes to decommission fully
> 5. Terminate the VMs the instances are running inside.
> A few notes. We did not shutdown the datanode processes, and the nodes were
> therefore not marked as dead by the namenode. We simply terminated the
> datanode VM (in this case an AWS instance). The nodes were marked as
> decommissioned. We are running our clusters with DNS, and when we terminate
> VMs, the associated CName is removed and no longer resolves. The errors do
> not seem to resolve without a restart.
> After we did this, the remaining regionservers started throwing
> NullPointerExceptions with the following stack trace:
> 2017-01-19 23:09:05,638 DEBUG org.apache.hadoop.hbase.ipc.RpcServer:
> RpcServer.RW.fifo.Q.read.handler=80,queue=14,port=60020: callId: 1727723891
> service: ClientService methodName: Scan size: 216 connection:
> 172.16.36.128:31538
> java.io.IOException
> at org.apache.hadoop.hbase.ipc.RpcServer.call(RpcServer.java:2214)
> at org.apache.hadoop.hbase.ipc.CallRunner.run(CallRunner.java:123)
> at
> org.apache.hadoop.hbase.ipc.RpcExecutor$Handler.run(RpcExecutor.java:204)
> at
> org.apache.hadoop.hbase.ipc.RpcExecutor$Handler.run(RpcExecutor.java:183)
> Caused by: java.lang.NullPointerException
> at org.apache.hadoop.hdfs.DFSInputStream.seek(DFSInputStream.java:1564)
> at org.apache.hadoop.fs.FSDataInputStream.seek(FSDataInputStream.java:62)
> at
> org.apache.hadoop.hbase.io.hfile.HFileBlock$AbstractFSReader.readAtOffset(HFileBlock.java:1434)
> at
> org.apache.hadoop.hbase.io.hfile.HFileBlock$FSReaderImpl.readBlockDataInternal(HFileBlock.java:1682)
> at
> org.apache.hadoop.hbase.io.hfile.HFileBlock$FSReaderImpl.readBlockData(HFileBlock.java:1542)
> at
> org.apache.hadoop.hbase.io.hfile.HFileReaderV2.readBlock(HFileReaderV2.java:445)
> at
> org.apache.hadoop.hbase.io.hfile.HFileBlockIndex$BlockIndexReader.loadDataBlockWithScanInfo(HFileBlockIndex.java:266)
> at
> org.apache.hadoop.hbase.io.hfile.HFileReaderV2$AbstractScannerV2.seekTo(HFileReaderV2.java:642)
> at
> org.apache.hadoop.hbase.io.hfile.HFileReaderV2$AbstractScannerV2.seekTo(HFileReaderV2.java:592)
> at
> org.apache.hadoop.hbase.regionserver.StoreFileScanner.seekAtOrAfter(StoreFileScanner.java:294)
> at
> org.apache.hadoop.hbase.regionserver.StoreFileScanner.seek(StoreFileScanner.java:199)
> at
> org.apache.hadoop.hbase.regionserver.StoreScanner.seekScanners(StoreScanner.java:343)
> at
> org.apache.hadoop.hbase.regionserver.StoreScanner.<init>(StoreScanner.java:198)
> at
> org.apache.hadoop.hbase.regionserver.HStore.createScanner(HStore.java:2106)
> at
> org.apache.hadoop.hbase.regionserver.HStore.getScanner(HStore.java:2096)
> at
> org.apache.hadoop.hbase.regionserver.HRegion$RegionScannerImpl.<init>(HRegion.java:5544)
> at
> org.apache.hadoop.hbase.regionserver.HRegion.instantiateRegionScanner(HRegion.java:2569)
> at
> org.apache.hadoop.hbase.regionserver.HRegion.getScanner(HRegion.java:2555)
> at
> org.apache.hadoop.hbase.regionserver.HRegion.getScanner(HRegion.java:2536)
> at
> org.apache.hadoop.hbase.regionserver.RSRpcServices.scan(RSRpcServices.java:2405)
> at
> org.apache.hadoop.hbase.protobuf.generated.ClientProtos$ClientService$2.callBlockingMethod(ClientProtos.java:33738)
> at org.apache.hadoop.hbase.ipc.RpcServer.call(RpcServer.java:2170)
> ... 3 more
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
