I'm trying to benchmark my application which uses Riak. At some point
riak_core_vnode processes message queues get overloaded. Here is the
detailed process_info information: http://kuku.jabber.ru/~xram/riak.log.
This leads to request timeouts on a client side.
Is it possible to tune Riak in order to avoid such situation?
I'm running Riak 1.3.1 (deb packet from basho.com) with mostly default
config. The difference is:
...
+ {riak_pipe, [
+ {worker_limit, 100},
+ {worker_queue_limit, 10000}
+ ]},
...
- {storage_backend, riak_kv_bitcask_backend},
+ {storage_backend, riak_kv_eleveldb_backend},
I'm using riakc (riak-erlang-client) with protobuffers as a client
library. I have the only one Riak node. Open files limit (ulimit -n) is
10000.
--
Regards,
Evgeniy Khramtsov, ProcessOne.
xmpp:[email protected].
_______________________________________________
riak-users mailing list
[email protected]
http://lists.basho.com/mailman/listinfo/riak-users_lists.basho.com
