I'd recommend lowering some of the leveldb configuration
defaults [1]. I'd start with the cache size. You'll notice that
most of these settings are _per_ _vnode_, which means
for your single-node cluster with 64 partitions (vnodes),
the defaults are likely a bit high for 10MM values.
It's also worth mentioning that Riak is not intended to be run
on a single node.
Reid
On Jul 29, 2012, at 2:01 PM, Antonio Ye wrote:
> this is on a single node and my ring_creation_size is 64.
>
> ring_num_partitions : 64
> ring_ownership : <<"[{'riak@127.0.0.1',64}]">>
> ring_creation_size : 64
>
> Thanks,
> Tony
>
>
> On Sun, Jul 29, 2012 at 9:12 AM, Reid Draper <reiddra...@gmail.com> wrote:
>> Hi Antonio,
>>
>> How many nodes do you have in your cluster? What is
>> the ring size ({ring_creation_size, N} in app.config)?
>> Are you seeing 3.6GB of memory used _per_ machine,
>> or total across the cluster?
>>
>> Reid
>>
>>
>> On Jul 29, 2012, at 2:37 AM, Antonio Ye wrote:
>>
>>> I am using Riak as the persistence data store for a distributed data
>>> grid so I really do not care to use Riak's or in my case eLevelDb's
>>> cache. The problem I have is that when I load 10 million keys into
>>> Riak it is taking 3.6g of memory. I have not set any of the eLevelDb
>>> backend configuration parameters therefore I expect Riak to use the
>>> defaults specified in the documentation. My bucket name is 7
>>> characters and my keys are 7 characters. Any ideas on how to make
>>> Riak/eLevelDb use as little memory as possible?
>>>
>>> _______________________________________________
>>> riak-users mailing list
>>> riak-users@lists.basho.com
>>> http://lists.basho.com/mailman/listinfo/riak-users_lists.basho.com
>>
_______________________________________________
riak-users mailing list
riak-users@lists.basho.com
http://lists.basho.com/mailman/listinfo/riak-users_lists.basho.com
