hbase-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Andrew Purtell (JIRA)" <j...@apache.org>
Subject [jira] Updated: (HBASE-2058) Coprocessors: CPU and memory limit policy enforcment via runtime weaving
Date Fri, 18 Dec 2009 21:55:18 GMT

     [ https://issues.apache.org/jira/browse/HBASE-2058?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Andrew Purtell updated HBASE-2058:
----------------------------------

    Description: 
Use the ASM bytecode analysis and rewriting engine to impose some constraints on CPU and memory
use. This is middle ground between arbitrary function and a locked down language.

We will be given arbitrary bytecode input. It is acceptable to reject a class and abort coprocessor
loading if it defies analysis at load time such that we have insufficient confidence about
the result.

Wrap allocations to simply disallow large allocations. Hook or add finalizers to keep a running
tally of aggregate heap charge. Disallow allocation beyond policy limit.

Weave CPU usage tracking into loop headers. Throw an uncatchable exception if time limits
prescribed by policy are exceeded.

> Coprocessors: CPU and memory limit policy enforcment via runtime weaving
> ------------------------------------------------------------------------
>
>                 Key: HBASE-2058
>                 URL: https://issues.apache.org/jira/browse/HBASE-2058
>             Project: Hadoop HBase
>          Issue Type: Sub-task
>            Reporter: Andrew Purtell
>
> Use the ASM bytecode analysis and rewriting engine to impose some constraints on CPU
and memory use. This is middle ground between arbitrary function and a locked down language.
> We will be given arbitrary bytecode input. It is acceptable to reject a class and abort
coprocessor loading if it defies analysis at load time such that we have insufficient confidence
about the result.
> Wrap allocations to simply disallow large allocations. Hook or add finalizers to keep
a running tally of aggregate heap charge. Disallow allocation beyond policy limit.
> Weave CPU usage tracking into loop headers. Throw an uncatchable exception if time limits
prescribed by policy are exceeded.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message