[ 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