[
https://issues.apache.org/jira/browse/HBASE-7205?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13502763#comment-13502763
]
Adrian Muraru commented on HBASE-7205:
--------------------------------------
[[email protected]] CHecking the thread classloader would be a quick fix in the
current implementation, however this would break the following usecase:
1. User creates a table and adds a RegionObserver CP (say
org.mycompany.MyRegionObserver) available in say hdfs:///lib/mycp-*0.1*.jar
(in turn, a new classloader will be cached on thread context)
2. The user updates the implementation and overwrite the COPROCESSOR att on
table to load from a new jar (say hdfs:///lib/mycp-*0.2*.jar)
In turn, the old/cached classloader will be used to load MyRegionObserver class
- in other ways there is no way to drop an old classloader
Though, this might be considered a corner case and documented properly
The real solution here in my mind would be to cache the classloaders keyed by
external jar canonical path:
This would require two new methods on org.apache.hadoop.hbase.Server
1. getExternalClassLoader(String key)/
2. setExternalClassLoader(String key, ClassLoader cl)
> Coprocessor classloader is replicated for all regions in the HRegionServer
> --------------------------------------------------------------------------
>
> Key: HBASE-7205
> URL: https://issues.apache.org/jira/browse/HBASE-7205
> Project: HBase
> Issue Type: Bug
> Components: Coprocessors
> Affects Versions: 0.92.2, 0.94.2
> Reporter: Adrian Muraru
> Priority: Critical
> Fix For: 0.96.0
>
>
> HBASE-6308 introduced a new custom CoprocessorClassLoader to load the
> coprocessor classes and a new instance of this CL is created for each single
> HRegion opened. This leads to OOME-PermGen when the number of regions go
> above hundres / region server.
> Having the table coprocessor jailed in a separate classloader is good however
> we should create only one for all regions of a table in each HRS.
--
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
