You are viewing a plain text version of this content. The canonical link for it is here.
Posted to hdfs-issues@hadoop.apache.org by "John Zhuge (JIRA)" <ji...@apache.org> on 2017/04/17 02:15:42 UTC
[jira] [Commented] (HDFS-11529) libHDFS still does not return
appropriate error information in many cases
[ https://issues.apache.org/jira/browse/HDFS-11529?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15970596#comment-15970596 ]
John Zhuge commented on HDFS-11529:
-----------------------------------
Thank you [~sailesh] for working on the patch. Here is my review for patch 002.
exception.c
#149: Why rename? Doesn’t the function still print it? This touches off so many one-line changes.
#178: Double space
exception.h
#23: Please update the javadoc:
expect.h
#168,169: Even though other places have done it, but I still find double-underscore prefix wrong because variables with double-underscore prefix are supposed to be reserved.
#168-181: Replace “str” and “substr” in macro body with “(str)” and “(substr)”
posix/thread_local_storage.c
#32: Any reason to remove the javadoc?
Testing
I tried the patch on Centos 7.2. Unit tests passed.
Any testing done on Windows?
> libHDFS still does not return appropriate error information in many cases
> -------------------------------------------------------------------------
>
> Key: HDFS-11529
> URL: https://issues.apache.org/jira/browse/HDFS-11529
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: libhdfs
> Affects Versions: 2.6.0
> Reporter: Sailesh Mukil
> Assignee: Sailesh Mukil
> Priority: Critical
> Labels: errorhandling, libhdfs
> Attachments: HDFS-11529.000.patch, HDFS-11529.001.patch, HDFS-11529.002.patch
>
>
> libHDFS uses a table to compare exceptions against and returns a corresponding error code to the application in case of an error.
> However, this table is manually populated and many times is disremembered when new exceptions are added.
> This causes libHDFS to return EINTERNAL (or Unknown Error(255)) whenever these exceptions are hit. These are some examples of exceptions that have been observed on an Error(255):
> org.apache.hadoop.ipc.StandbyException (Operation category WRITE is not supported in state standby)
> java.io.EOFException: Cannot seek after EOF
> javax.security.sasl.SaslException: GSS initiate failed [Caused by GSSException: No valid credentials provided (Mechanism level: Failed to find any Kerberos tgt)
> It is of course not possible to have an error code for each and every type of exception, so one suggestion of how this can be addressed is by having a call such as hdfsGetLastException() that would return the last exception that a libHDFS thread encountered. This way, an application may choose to call hdfsGetLastException() if it receives EINTERNAL.
> We can make use of the Thread Local Storage to store this information. Also, this makes sure that the current functionality is preserved.
> This is a follow up from HDFS-4997.
--
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