[
https://issues.apache.org/jira/browse/HDFS-9668?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15197893#comment-15197893
]
Colin Patrick McCabe commented on HDFS-9668:
--------------------------------------------
Arpit wrote:
bq. Detecting slow disks is a known problem for DataNodes. If this problem does
not manifest in regular operation perhaps we should try to add slow disk
detection instead.
There are a lot of reasons why a disk I/O operation could block. Many of them
have nothing to do with the disk being bad. For example, on Linux, the value
of {{/proc/sys/vm/dirty_writeback_centisecs}} controls how long we will wait
before writing data that is in the page cache to disk. This value exists so
that Linux does not accumulate a lot of cached modifications that haven't been
written to disk and then lose them all when the power goes out. Of course, if
the disk is busy writing something that was aged out of the page cache, that
means that your {{createRbw}} operation may have to "get in line." A typical
hard can only do around 100 seeks a second.
This is why disk I/O is a blocking operation in practice, even with very fast
disks. It is simply not a good idea to hold a lot of locks while doing disk
I/O, because you will end up blocking threads that don't care (and shouldn't
have to care) about the I/O you are doing.
Nicholas wrote:
bq. One issue concern me about this is: how could we make sure that the
synchronization is correct, especially outside the class? The current patch
only changes FsDatasetImpl but the FsDatasetImpl object is also synchronized in
other classes such as FsVolumeImpl.
Yes, we should also look at other places that are locking FsDatasetImpl
externally (and hopefully change those places so they don't have to do that?)
> Optimize the locking in FsDatasetImpl
> -------------------------------------
>
> Key: HDFS-9668
> URL: https://issues.apache.org/jira/browse/HDFS-9668
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: datanode
> Reporter: Jingcheng Du
> Assignee: Jingcheng Du
> Attachments: HDFS-9668-1.patch, HDFS-9668-2.patch, execution_time.png
>
>
> During the HBase test on a tiered storage of HDFS (WAL is stored in
> SSD/RAMDISK, and all other files are stored in HDD), we observe many
> long-time BLOCKED threads on FsDatasetImpl in DataNode. The following is part
> of the jstack result:
> {noformat}
> "DataXceiver for client DFSClient_NONMAPREDUCE_-1626037897_1 at
> /192.168.50.16:48521 [Receiving block
> BP-1042877462-192.168.50.13-1446173170517:blk_1073779272_40852]" - Thread
> t@93336
> java.lang.Thread.State: BLOCKED
> at
> org.apache.hadoop.hdfs.server.datanode.fsdataset.impl.FsDatasetImpl.createRbw(FsDatasetImpl.java:1111)
> - waiting to lock <18324c9> (a
> org.apache.hadoop.hdfs.server.datanode.fsdataset.impl.FsDatasetImpl) owned by
> "DataXceiver for client DFSClient_NONMAPREDUCE_-1626037897_1 at
> /192.168.50.16:48520 [Receiving block
> BP-1042877462-192.168.50.13-1446173170517:blk_1073779271_40851]" t@93335
> at
> org.apache.hadoop.hdfs.server.datanode.fsdataset.impl.FsDatasetImpl.createRbw(FsDatasetImpl.java:113)
> at
> org.apache.hadoop.hdfs.server.datanode.BlockReceiver.<init>(BlockReceiver.java:183)
> at
> org.apache.hadoop.hdfs.server.datanode.DataXceiver.writeBlock(DataXceiver.java:615)
> at
> org.apache.hadoop.hdfs.protocol.datatransfer.Receiver.opWriteBlock(Receiver.java:137)
> at
> org.apache.hadoop.hdfs.protocol.datatransfer.Receiver.processOp(Receiver.java:74)
> at
> org.apache.hadoop.hdfs.server.datanode.DataXceiver.run(DataXceiver.java:235)
> at java.lang.Thread.run(Thread.java:745)
> Locked ownable synchronizers:
> - None
>
> "DataXceiver for client DFSClient_NONMAPREDUCE_-1626037897_1 at
> /192.168.50.16:48520 [Receiving block
> BP-1042877462-192.168.50.13-1446173170517:blk_1073779271_40851]" - Thread
> t@93335
> java.lang.Thread.State: RUNNABLE
> at java.io.UnixFileSystem.createFileExclusively(Native Method)
> at java.io.File.createNewFile(File.java:1012)
> at
> org.apache.hadoop.hdfs.server.datanode.DatanodeUtil.createTmpFile(DatanodeUtil.java:66)
> at
> org.apache.hadoop.hdfs.server.datanode.fsdataset.impl.BlockPoolSlice.createRbwFile(BlockPoolSlice.java:271)
> at
> org.apache.hadoop.hdfs.server.datanode.fsdataset.impl.FsVolumeImpl.createRbwFile(FsVolumeImpl.java:286)
> at
> org.apache.hadoop.hdfs.server.datanode.fsdataset.impl.FsDatasetImpl.createRbw(FsDatasetImpl.java:1140)
> - locked <18324c9> (a
> org.apache.hadoop.hdfs.server.datanode.fsdataset.impl.FsDatasetImpl)
> at
> org.apache.hadoop.hdfs.server.datanode.fsdataset.impl.FsDatasetImpl.createRbw(FsDatasetImpl.java:113)
> at
> org.apache.hadoop.hdfs.server.datanode.BlockReceiver.<init>(BlockReceiver.java:183)
> at
> org.apache.hadoop.hdfs.server.datanode.DataXceiver.writeBlock(DataXceiver.java:615)
> at
> org.apache.hadoop.hdfs.protocol.datatransfer.Receiver.opWriteBlock(Receiver.java:137)
> at
> org.apache.hadoop.hdfs.protocol.datatransfer.Receiver.processOp(Receiver.java:74)
> at
> org.apache.hadoop.hdfs.server.datanode.DataXceiver.run(DataXceiver.java:235)
> at java.lang.Thread.run(Thread.java:745)
> Locked ownable synchronizers:
> - None
> {noformat}
> We measured the execution of some operations in FsDatasetImpl during the
> test. Here following is the result.
> !execution_time.png!
> The operations of finalizeBlock, addBlock and createRbw on HDD in a heavy
> load take a really long time.
> It means one slow operation of finalizeBlock, addBlock and createRbw in a
> slow storage can block all the other same operations in the same DataNode,
> especially in HBase when many wal/flusher/compactor are configured.
> We need a finer grained lock mechanism in a new FsDatasetImpl implementation
> and users can choose the implementation by configuring
> "dfs.datanode.fsdataset.factory" in DataNode.
> We can implement the lock by either storage level or block-level.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
