You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hbase.apache.org by "Jiuming Huang (JIRA)" <ji...@apache.org> on 2012/11/06 04:12:11 UTC
[jira] [Created] (HBASE-7102) The process of disappear without any
exception
Jiuming Huang created HBASE-7102:
------------------------------------
Summary: The process of disappear without any exception
Key: HBASE-7102
URL: https://issues.apache.org/jira/browse/HBASE-7102
Project: HBase
Issue Type: Bug
Components: regionserver
Affects Versions: 0.94.2
Environment: My HBase culster contains 7 computers. One is the master and zookeeper server, the other 6 are the region servers.
The operatiing system is Centos 5.6 with kenerl 2.6.18-238.el5.
The jdk version is 1.7.0_03
Reporter: Jiuming Huang
The culster is started normally. It can work when the I/O pressure is small.
However, when I run a large pressure job (with a lot of threads, each of which writes and reads frequently) about one hour, one of the region server will crash.
I investigated the logs of the HRegionServer, they didn't contain any exception log.
Specifically, the log of the down server ends with some normal log (info level log).
This bug can be replayed easily, and each time the crashed server is different. Even more, the log of the crashed server ends with different information for different experiment.
I really don't know why the process of HRegionServer disappear so weirdly.
If the process is crashed due to my configuration or the enviornment, the log should contain some exception information, right?
So, I doubt the reason is the JVM crashed. But I didn't find any error log in JVM.
How to go on the test to find the reason?
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (HBASE-7102) The process of disappear without
any exception
Posted by "Ted Yu (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-7102?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13491218#comment-13491218 ]
Ted Yu commented on HBASE-7102:
-------------------------------
How many regions did one region server serve ?
Can you enable the following so that we can look at heap nump ?
{code}
JAVA_OPTS=-XX:-HeapDumpOnOutOfMemoryError
JAVA_OPTS=-XX:HeapDumpPath=/var/dump
{code}
Normally people start describing the problems they encountered on dev@hbase.apache.org
JIRA is opened after we have some idea about the root cause.
Thanks
> The process of disappear without any exception
> ----------------------------------------------
>
> Key: HBASE-7102
> URL: https://issues.apache.org/jira/browse/HBASE-7102
> Project: HBase
> Issue Type: Bug
> Components: regionserver
> Affects Versions: 0.94.2
> Environment: My HBase culster contains 7 computers. One is the master and zookeeper server, the other 6 are the region servers.
> The operatiing system is Centos 5.6 with kenerl 2.6.18-238.el5.
> The jdk version is 1.7.0_03
> Reporter: Jiuming Huang
>
> The culster is started normally. It can work when the I/O pressure is small.
> However, when I run a large pressure job (with a lot of threads, each of which writes and reads frequently) about one hour, one of the region server will crash.
> I investigated the logs of the HRegionServer, they didn't contain any exception log.
> Specifically, the log of the down server ends with some normal log (info level log).
> This bug can be replayed easily, and each time the crashed server is different. Even more, the log of the crashed server ends with different information for different experiment.
> I really don't know why the process of HRegionServer disappear so weirdly.
> If the process is crashed due to my configuration or the enviornment, the log should contain some exception information, right?
> So, I doubt the reason is the JVM crashed. But I didn't find any error log in JVM.
> How to go on the test to find the reason?
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Resolved] (HBASE-7102) The process of disappear without any
exception
Posted by "stack (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HBASE-7102?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
stack resolved HBASE-7102.
--------------------------
Resolution: Invalid
What Ted said. This is for mailing list (Did you check the .out file? You are using a very early 1.7 version. Even the oracle fellas say you shouldn't use such early versions of 1.7.
> The process of disappear without any exception
> ----------------------------------------------
>
> Key: HBASE-7102
> URL: https://issues.apache.org/jira/browse/HBASE-7102
> Project: HBase
> Issue Type: Bug
> Components: regionserver
> Affects Versions: 0.94.2
> Environment: My HBase culster contains 7 computers. One is the master and zookeeper server, the other 6 are the region servers.
> The operatiing system is Centos 5.6 with kenerl 2.6.18-238.el5.
> The jdk version is 1.7.0_03
> Reporter: Jiuming Huang
>
> The culster is started normally. It can work when the I/O pressure is small.
> However, when I run a large pressure job (with a lot of threads, each of which writes and reads frequently) about one hour, one of the region server will crash.
> I investigated the logs of the HRegionServer, they didn't contain any exception log.
> Specifically, the log of the down server ends with some normal log (info level log).
> This bug can be replayed easily, and each time the crashed server is different. Even more, the log of the crashed server ends with different information for different experiment.
> I really don't know why the process of HRegionServer disappear so weirdly.
> If the process is crashed due to my configuration or the enviornment, the log should contain some exception information, right?
> So, I doubt the reason is the JVM crashed. But I didn't find any error log in JVM.
> How to go on the test to find the reason?
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira