[
https://issues.apache.org/jira/browse/HDFS-7060?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14391501#comment-14391501
]
Colin Patrick McCabe edited comment on HDFS-7060 at 4/1/15 9:31 PM:
--------------------------------------------------------------------
While I agree that making the DataNode heartbeat lockless might be a good idea
in theory, it is not a bugfix. It is an optimization or improvement.
In contrast, the issue with {{FsDatasetImpl#createTemporary}} holding the lock
for a long period is a bug, and it needs to be fixed. Even if the heartbeat is
made lockless, HDFS-7999 will still cause many problems such as extremely high
write and read latency, unresponsive datanodes, and so on. Making the
heartbeat lockless does not fix HDFS-7999, although it might alleviate a few of
the symptoms.
Edit: I see that Haohui explained why we might be able to remove some of these
locks. I'll try to take a look later
was (Author: cmccabe):
While I agree that making the DataNode heartbeat lockless might be a good idea
in theory, it is not a bugfix. It is an optimization or improvement.
In contrast, the issue with {{FsDatasetImpl#createTemporary}} holding the lock
for a long period is a bug, and it needs to be fixed. Even if the heartbeat is
made lockless, HDFS-7999 will still cause many problems such as extremely high
write and read latency, unresponsive datanodes, and so on. Making the
heartbeat lockless does not fix HDFS-7999, although it might alleviate a few of
the symptoms.
The patch here seems to remove a bunch of synchronized blocks, but I don't see
any explanation for how we will avoid race conditions associated with
concurrent access that these synchronized blocks were protecting against. This
needs to be be much better thought-out before we can consider it.
> Contentions of the monitor of FsDatasetImpl block DN's heartbeat
> ----------------------------------------------------------------
>
> Key: HDFS-7060
> URL: https://issues.apache.org/jira/browse/HDFS-7060
> Project: Hadoop HDFS
> Issue Type: Bug
> Reporter: Haohui Mai
> Assignee: Xinwei Qin
> Priority: Critical
> Attachments: HDFS-7060-002.patch, HDFS-7060.000.patch,
> HDFS-7060.001.patch
>
>
> We're seeing the heartbeat is blocked by the monitor of {{FsDatasetImpl}}
> when the DN is under heavy load of writes:
> {noformat}
> java.lang.Thread.State: BLOCKED (on object monitor)
> at
> org.apache.hadoop.hdfs.server.datanode.fsdataset.impl.FsVolumeImpl.getDfsUsed(FsVolumeImpl.java:115)
> - waiting to lock <0x0000000780304fb8> (a
> org.apache.hadoop.hdfs.server.datanode.fsdataset.impl.FsDatasetImpl)
> at
> org.apache.hadoop.hdfs.server.datanode.fsdataset.impl.FsDatasetImpl.getStorageReports(FsDatasetImpl.java:91)
> - locked <0x0000000780612fd8> (a java.lang.Object)
> at
> org.apache.hadoop.hdfs.server.datanode.BPServiceActor.sendHeartBeat(BPServiceActor.java:563)
> at
> org.apache.hadoop.hdfs.server.datanode.BPServiceActor.offerService(BPServiceActor.java:668)
> at
> org.apache.hadoop.hdfs.server.datanode.BPServiceActor.run(BPServiceActor.java:827)
> at java.lang.Thread.run(Thread.java:744)
> java.lang.Thread.State: BLOCKED (on object monitor)
> at
> org.apache.hadoop.hdfs.server.datanode.fsdataset.impl.FsDatasetImpl.createRbw(FsDatasetImpl.java:743)
> - waiting to lock <0x0000000780304fb8> (a
> org.apache.hadoop.hdfs.server.datanode.fsdataset.impl.FsDatasetImpl)
> at
> org.apache.hadoop.hdfs.server.datanode.fsdataset.impl.FsDatasetImpl.createRbw(FsDatasetImpl.java:60)
> at
> org.apache.hadoop.hdfs.server.datanode.BlockReceiver.<init>(BlockReceiver.java:169)
> at
> org.apache.hadoop.hdfs.server.datanode.DataXceiver.writeBlock(DataXceiver.java:621)
> at
> org.apache.hadoop.hdfs.protocol.datatransfer.Receiver.opWriteBlock(Receiver.java:124)
> at
> org.apache.hadoop.hdfs.protocol.datatransfer.Receiver.processOp(Receiver.java:71)
> at
> org.apache.hadoop.hdfs.server.datanode.DataXceiver.run(DataXceiver.java:232)
> at java.lang.Thread.run(Thread.java:744)
> java.lang.Thread.State: RUNNABLE
> at java.io.UnixFileSystem.createFileExclusively(Native Method)
> at java.io.File.createNewFile(File.java:1006)
> at
> org.apache.hadoop.hdfs.server.datanode.DatanodeUtil.createTmpFile(DatanodeUtil.java:59)
> at
> org.apache.hadoop.hdfs.server.datanode.fsdataset.impl.BlockPoolSlice.createRbwFile(BlockPoolSlice.java:244)
> at
> org.apache.hadoop.hdfs.server.datanode.fsdataset.impl.FsVolumeImpl.createRbwFile(FsVolumeImpl.java:195)
> at
> org.apache.hadoop.hdfs.server.datanode.fsdataset.impl.FsDatasetImpl.createRbw(FsDatasetImpl.java:753)
> - locked <0x0000000780304fb8> (a
> org.apache.hadoop.hdfs.server.datanode.fsdataset.impl.FsDatasetImpl)
> at
> org.apache.hadoop.hdfs.server.datanode.fsdataset.impl.FsDatasetImpl.createRbw(FsDatasetImpl.java:60)
> at
> org.apache.hadoop.hdfs.server.datanode.BlockReceiver.<init>(BlockReceiver.java:169)
> at
> org.apache.hadoop.hdfs.server.datanode.DataXceiver.writeBlock(DataXceiver.java:621)
> at
> org.apache.hadoop.hdfs.protocol.datatransfer.Receiver.opWriteBlock(Receiver.java:124)
> at
> org.apache.hadoop.hdfs.protocol.datatransfer.Receiver.processOp(Receiver.java:71)
> at
> org.apache.hadoop.hdfs.server.datanode.DataXceiver.run(DataXceiver.java:232)
> at java.lang.Thread.run(Thread.java:744)
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
