cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jonathan Ellis <jbel...@gmail.com>
Subject Re: Native heap leaks?
Date Fri, 06 May 2011 16:07:54 GMT
On Thu, May 5, 2011 at 6:16 PM, aaron morton <aaron@thelastpickle.com> wrote:
> Hannes,
>        To get a baseline of behaviour set disk_access to standard. You will probably
want to keep it like that if you want better control over the memory on the box.

Changing disk access mode just gives you the illusion of more control;
the same buffering is happening except now you force Cassandra to do
extra copies: http://wiki.apache.org/cassandra/FAQ#mmap

The ONLY reason non-mmap'd i/o is still around is to support testing
on 32bit machines.

-- 
Jonathan Ellis
Project Chair, Apache Cassandra
co-founder of DataStax, the source for professional Cassandra support
http://www.datastax.com

Mime
View raw message