hadoop-hdfs-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Yongtao Yang (JIRA)" <j...@apache.org>
Subject [jira] [Comment Edited] (HDFS-8409) HDFS client RPC call throws "java.lang.IllegalStateException"
Date Fri, 03 Mar 2017 07:12:45 GMT

    [ https://issues.apache.org/jira/browse/HDFS-8409?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15892062#comment-15892062
] 

Yongtao Yang edited comment on HDFS-8409 at 3/3/17 7:12 AM:
------------------------------------------------------------

I found a clue that may be useful to this problem. {{DFSClient#namenode}} is a proxy whose
{{InvocationHandler}} is {{RetryInvocationHandler}}. If {{namenode.toString()}} or {{namenode.hashCode()}}
or some other non-rpc methods are called, then the next real rpc method will failed at {{setCallIdAndRetryCount()}}.
The cause of this is that {{toString()}} is also forwarded to {{RetryInvocationHandler.invoke()}}
and {{setCallIdAndRetryCount()}} is executed there, then {{ipc.Client#callId.get()}} will
not be {{null}} any more. For real rpc methods, {{ipc.Client#callId}} will be reset to the
value {{null}} when creating a {{Call}} instance({{org.apache.hadoop.ipc.Client.Call.Call(RpcKind,
Writable)}}), but for {{toString()}} or {{hashCode()}}, it can't reach the {{Call}} constructor,
so {{ipc.Client#callId}} will never recover to be {{null}} and the next method(whether it
is a real rpc or not) will fail.


was (Author: ytyang):
I found a clue that may be useful to this problem. {{DFSClient#namenode}} is a proxy whose
{{InvocationHandler}} is {{RetryInvocationHandler}}. If {{namenode.toString()}} or {{namenode.hashCode()}}
or some other non-rpc methods are called, then the next real rpc method will failed at {{setCallIdAndRetryCount()}}.
The cause of this is that {{toString()}} is also forwarded to {{RetryInvocationHandler.invoke()}}
and {{setCallIdAndRetryCount()}} is executed there, then {{ipc.Client#callId.get()}} will
not be null any more. For real rpc methods, {{ipc.Client#callId}} will be reset to null when
creating a {{Call}} instance({{org.apache.hadoop.ipc.Client.Call.Call(RpcKind, Writable)}}),
but for {{toString()}} or {{hashCode()}}, it can't reach the {{Call}} constructor, so {{ipc.Client#callId}}
will never recover to be null and the next method(whether it is a real rpc or not) will fail.

> HDFS client RPC call throws "java.lang.IllegalStateException"
> -------------------------------------------------------------
>
>                 Key: HDFS-8409
>                 URL: https://issues.apache.org/jira/browse/HDFS-8409
>             Project: Hadoop HDFS
>          Issue Type: Bug
>          Components: hdfs-client
>            Reporter: Juan Yu
>            Assignee: Juan Yu
>         Attachments: HDFS-8409.001.patch, HDFS-8409.002.patch, HDFS-8409.003.patch
>
>
> When the HDFS client RPC calls need to retry, it sometimes throws "java.lang.IllegalStateException"
and retry is aborted and cause the client call will fail.
> {code}
> Caused by: java.lang.IllegalStateException
> 	at com.google.common.base.Preconditions.checkState(Preconditions.java:129)
> 	at org.apache.hadoop.ipc.Client.setCallIdAndRetryCount(Client.java:116)
> 	at org.apache.hadoop.io.retry.RetryInvocationHandler.invoke(RetryInvocationHandler.java:99)
> 	at com.sun.proxy.$Proxy16.getFileInfo(Unknown Source)
> 	at org.apache.hadoop.hdfs.DFSClient.getFileInfo(DFSClient.java:1912)
> 	at org.apache.hadoop.hdfs.DistributedFileSystem$17.doCall(DistributedFileSystem.java:1089)
> 	at org.apache.hadoop.hdfs.DistributedFileSystem$17.doCall(DistributedFileSystem.java:1085)
> 	at org.apache.hadoop.fs.FileSystemLinkResolver.resolve(FileSystemLinkResolver.java:81)
> 	at org.apache.hadoop.hdfs.DistributedFileSystem.getFileStatus(DistributedFileSystem.java:1085)
> 	at org.apache.hadoop.fs.FileSystem.exists(FileSystem.java:1400)
> {code}
> Here is the check that throws exception
> {code}
>   public static void setCallIdAndRetryCount(int cid, int rc) {
>       ...
>       Preconditions.checkState(callId.get() == null);
>   }
> {code}
> The RetryInvocationHandler tries to call it with not null callId and causes exception.



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)

---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-help@hadoop.apache.org


Mime
View raw message