[
https://issues.apache.org/jira/browse/HDFS-14646?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Xudong Cao updated HDFS-14646:
------------------------------
Comment: was deleted
(was: *Test Result:*
in a 3 nodes HDFS: ubuntu1 (ANN) +ubuntu2 (SNN) + ubuntu3(SNN), the uploading
log in ubuntu2 and ubuntu3 is as follows:
1. SNN ubuntu2:
{code:java}
root@ubuntu2:~/hadoop-3.3.0-SNAPSHOT/logs# grep "Uploaded"
hadoop-root-namenode-ubuntu2.log
2019-07-16 01:52:24,801 INFO
org.apache.hadoop.hdfs.server.namenode.TransferFsImage: Uploaded image with
txid 9573 to namenode at http://ubuntu1:9870 in 0.178 seconds
2019-07-16 01:53:24,912 INFO
org.apache.hadoop.hdfs.server.namenode.TransferFsImage: Uploaded image with
txid 9759 to namenode at http://ubuntu1:9870 in 0.041 seconds
2019-07-16 01:54:25,051 INFO
org.apache.hadoop.hdfs.server.namenode.TransferFsImage: Uploaded image with
txid 9777 to namenode at http://ubuntu1:9870 in 0.075 seconds
2019-07-16 01:55:25,147 INFO
org.apache.hadoop.hdfs.server.namenode.TransferFsImage: Uploaded image with
txid 9961 to namenode at http://ubuntu1:9870 in 0.031 seconds
2019-07-16 01:56:25,253 INFO
org.apache.hadoop.hdfs.server.namenode.TransferFsImage: Uploaded image with
txid 9981 to namenode at http://ubuntu1:9870 in 0.054 seconds
2019-07-16 01:57:25,323 INFO
org.apache.hadoop.hdfs.server.namenode.TransferFsImage: Uploaded image with
txid 10171 to namenode at http://ubuntu1:9870 in 0.033 seconds
2019-07-16 01:58:25,388 INFO
org.apache.hadoop.hdfs.server.namenode.TransferFsImage: Uploaded image with
txid 10191 to namenode at http://ubuntu1:9870 in 0.032 seconds
2019-07-16 01:59:25,479 INFO
org.apache.hadoop.hdfs.server.namenode.TransferFsImage: Uploaded image with
txid 10383 to namenode at http://ubuntu1:9870 in 0.046 seconds{code}
2. another SNN ubuntu3:
{code:java}
root@ubuntu3:~/hadoop-3.3.0-SNAPSHOT/logs# grep "Uploaded"
hadoop-root-namenode-ubuntu3.log
2019-07-16 02:00:34,767 INFO
org.apache.hadoop.hdfs.server.namenode.TransferFsImage: Uploaded image with
txid 10401 to namenode at http://ubuntu1:9870 in 0.028 seconds
2019-07-16 02:02:34,851 INFO
org.apache.hadoop.hdfs.server.namenode.TransferFsImage: Uploaded image with
txid 10603 to namenode at http://ubuntu1:9870 in 0.03 seconds
2019-07-16 02:04:34,938 INFO
org.apache.hadoop.hdfs.server.namenode.TransferFsImage: Uploaded image with
txid 10807 to namenode at http://ubuntu1:9870 in 0.033 seconds
2019-07-16 02:06:35,021 INFO
org.apache.hadoop.hdfs.server.namenode.TransferFsImage: Uploaded image with
txid 11013 to namenode at http://ubuntu1:9870 in 0.041 seconds
2019-07-16 02:08:35,094 INFO
org.apache.hadoop.hdfs.server.namenode.TransferFsImage: Uploaded image with
txid 11217 to namenode at http://ubuntu1:9870 in 0.029 seconds
2019-07-16 02:10:35,200 INFO
org.apache.hadoop.hdfs.server.namenode.TransferFsImage: Uploaded image with
txid 11423 to namenode at http://ubuntu1:9870 in 0.032 seconds
2019-07-16 02:12:35,285 INFO
org.apache.hadoop.hdfs.server.namenode.TransferFsImage: Uploaded image with
txid 11629 to namenode at http://ubuntu1:9870 in 0.026 seconds
2019-07-16 02:14:35,357 INFO
org.apache.hadoop.hdfs.server.namenode.TransferFsImage: Uploaded image with
txid 11835 to namenode at http://ubuntu1:9870 in 0.023 seconds
2019-07-16 02:16:35,442 INFO
org.apache.hadoop.hdfs.server.namenode.TransferFsImage: Uploaded image with
txid 12035 to namenode at http://ubuntu1:9870 in 0.042 seconds
2019-07-16 02:18:35,515 INFO
org.apache.hadoop.hdfs.server.namenode.TransferFsImage: Uploaded image with
txid 12233 to namenode at http://ubuntu1:9870 in 0.031 seconds
2019-07-16 02:20:35,605 INFO
org.apache.hadoop.hdfs.server.namenode.TransferFsImage: Uploaded image with
txid 12441 to namenode at http://ubuntu1:9870 in 0.033 seconds
2019-07-16 02:22:35,675 INFO
org.apache.hadoop.hdfs.server.namenode.TransferFsImage: Uploaded image with
txid 12647 to namenode at http://ubuntu1:9870 in 0.029 seconds
2019-07-16 02:24:35,771 INFO
org.apache.hadoop.hdfs.server.namenode.TransferFsImage: Uploaded image with
txid 12853 to namenode at http://ubuntu1:9870 in 0.041 seconds{code})
> Standby NameNode should not upload fsimage to an inappropriate NameNode.
> ------------------------------------------------------------------------
>
> Key: HDFS-14646
> URL: https://issues.apache.org/jira/browse/HDFS-14646
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: hdfs
> Affects Versions: 3.1.2
> Reporter: Xudong Cao
> Assignee: Xudong Cao
> Priority: Major
> Attachments: HDFS-14646.000.patch
>
>
> *Problem Description:*
> In the multi-NameNode scenario, when a SNN uploads a FsImage, it will put
> the image to all other NNs (whether the peer NN is an ANN or not), and even
> if the peer NN immediately replies with an error (such as
> TransferResult.NOT_ACTIVE_NAMENODE_FAILURE, TransferResult
> .OLD_TRANSACTION_ID_FAILURE, etc.), the local SNN will not terminate the put
> process immediately, but will put the FsImage completely to the peer NN, and
> will not read the peer NN's reply until the put is completed.
> Depending on the version of Jetty, this behavior can lead to different
> consequences, I tested it under 2.7.2 and the latest 3.3.0 version.
> *1. In Hadoop 2.7.2 (with Jetty 6.1.26)*
> After perr NN called HttpServletResponse.sendError(), the underlying TCP
> connection will still be established, and the data SNN sent will be read by
> Jetty framework itself in the peer NN, so the SNN will still insignificantly
> send the FsImage to the peer NN continuously, causing a waste of time and
> bandwidth. In a relatively large HDFS cluster, the size of FsImage can often
> reach about 30GB, This is indeed a big waste.
> *2.In newest Hadoop-3.3.0-SNAPSHOT (with Jetty 9.3.27)*
> After perr NN called HttpServletResponse.sendError(), the underlying TCP
> connection will be auto closed, and then SNN will directly get an "Error
> writing request body to server" exception, as below:
> {code:java}
> 2019-08-17 03:59:25,413 INFO namenode.TransferFsImage: Sending fileName:
> /tmp/hadoop-root/dfs/name/current/fsimage_0000000000003364240, fileSize:
> 9864721. Sent total: 524288 bytes. Size of last segment intended to send:
> 4096 bytes.
> java.io.IOException: Error writing request body to server
> at
> sun.net.www.protocol.http.HttpURLConnection$StreamingOutputStream.checkError(HttpURLConnection.java:3587)
> at
> sun.net.www.protocol.http.HttpURLConnection$StreamingOutputStream.write(HttpURLConnection.java:3570)
> at
> org.apache.hadoop.hdfs.server.namenode.TransferFsImage.copyFileToStream(TransferFsImage.java:396)
> at
> org.apache.hadoop.hdfs.server.namenode.TransferFsImage.writeFileToPutRequest(TransferFsImage.java:340)
> at
> org.apache.hadoop.hdfs.server.namenode.TransferFsImage.uploadImage(TransferFsImage.java:314)
> at
> org.apache.hadoop.hdfs.server.namenode.TransferFsImage.uploadImageFromStorage(TransferFsImage.java:249)
> at
> org.apache.hadoop.hdfs.server.namenode.ha.StandbyCheckpointer$1.call(StandbyCheckpointer.java:277)
> at
> org.apache.hadoop.hdfs.server.namenode.ha.StandbyCheckpointer$1.call(StandbyCheckpointer.java:272)
> at java.util.concurrent.FutureTask.run(FutureTask.java:266)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
> at java.lang.Thread.run(Thread.java:748)
> 2019-08-17 03:59:25,422 INFO namenode.TransferFsImage: Sending fileName:
> /tmp/hadoop-root/dfs/name/current/fsimage_0000000000003364240, fileSize:
> 9864721. Sent total: 851968 bytes. Size of last segment intended to send:
> 4096 bytes.
> java.io.IOException: Error writing request body to server
> at
> sun.net.www.protocol.http.HttpURLConnection$StreamingOutputStream.checkError(HttpURLConnection.java:3587)
> at
> sun.net.www.protocol.http.HttpURLConnection$StreamingOutputStream.write(HttpURLConnection.java:3570)
> at
> org.apache.hadoop.hdfs.server.namenode.TransferFsImage.copyFileToStream(TransferFsImage.java:396)
> at
> org.apache.hadoop.hdfs.server.namenode.TransferFsImage.writeFileToPutRequest(TransferFsImage.java:340)
> {code}
>
> *Solution:*
> A standby NameNode should not upload fsimage to an inappropriate NameNode,
> when the he plans to put a FsImage to the peer NN, he need to check whether
> he really need to put it at this time.
--
This message was sent by Atlassian JIRA
(v7.6.14#76016)
---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
