[ 
https://issues.apache.org/jira/browse/HBASE-4391?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

stack updated HBASE-4391:
-------------------------

    Release Note: 
A new profile, native, has been added to pom.xml  Adding '-Pnative' will build 
libmlockall_agent.so   HBase will look for the .so under  ./lib/native/PLATFORM 
(See JAVA_LIBRARY_PATH in the bin/hbase starup script).

mlockall_agent is a simple VM Agent that allows to lock the address space of 
the region server process. This avoids the process' memory eviction under 
pressure.

You can load the agent by adding it as a jvm option:
 export HBASE_REGIONSERVER_OPTS="-agentpath:./libmlockall_agent.so=user=hbase"

To use this feature, execute the following command as root:
hbase --mlock user=hbase regionserver start

  was:
A new profile, native, has been added to pom.xml  Adding '-Pnative' will build 
libmlockall_agent.so

mlockall_agent is a simple VM Agent that allows to lock the address space of 
the region server process. This avoids the process' memory eviction under 
pressure.

You can load the agent by adding it as a jvm option:
 export HBASE_REGIONSERVER_OPTS="-agentpath:./libmlockall_agent.so=user=hbase"

To use this feature, execute the following command as root:
hbase --mlock user=hbase regionserver start

    
> Add ability to start RS as root and call mlockall
> -------------------------------------------------
>
>                 Key: HBASE-4391
>                 URL: https://issues.apache.org/jira/browse/HBASE-4391
>             Project: HBase
>          Issue Type: New Feature
>          Components: regionserver
>    Affects Versions: 0.94.0
>            Reporter: Todd Lipcon
>            Assignee: Matteo Bertozzi
>             Fix For: 0.95.0
>
>         Attachments: 4391-v4.patch, HBASE-4391-v0.patch, HBASE-4391-v1.patch, 
> HBASE-4391-v2.patch, HBASE-4391-v3.patch
>
>
> A common issue we've seen in practice is that users oversubscribe their 
> region servers with too many MR tasks, etc. As soon as the machine starts 
> swapping, the RS grinds to a halt, loses ZK session, aborts, etc.
> This can be combatted by starting the RS as root, calling mlockall(), and 
> then setuid down to the hbase user. We should not require this, but we should 
> provide it as an option.

--
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

Reply via email to