Riak 0.11.0 hanging on OSX 10.6

2010-07-18 Thread Shane Holloway
I'm having a bit of a problem with Riak.  When I keep my new documents to about 
10 per second, Riak works fine.  But as I increase my speed to about 100 docs 
per second, Riak gets dramatically slow.  After a while, the beam process takes 
100% of my CPU and no longer responds to ping/stop or any other command from 
the bin folder.  I end up sending kill TERM to stop the process.  

If I restart Riak, the bitcask backend seems to be limited to about 1-2 
document stores per second, but read seems fairly responsive still.  I'd 
appreciate any thoughts or troubleshooting tips you may have.  It's an 
in-development database with demo data, so I can run any experiments you'd like 
to see.

Thanks,
-Shane



Riak/riakOne% ./bin/riak-admin status
1-minute stats for 'r...@172.20.0.49'
---
vnode gets : 104
vnode_puts : 103
vnode_gets_total : 136
vnode_puts_total : 127
node_gets : 36
node_gets_total : 47
node_get_fsm_time_mean : 2635.3
node_get_fsm_time_median : 667
node_get_fsm_time_95 : 9156
node_get_fsm_time_99 : 40652
node_get_fsm_time_100 : 40652
node_puts : 36
node_puts_total : 44
node_put_fsm_time_mean : 1670575.6
node_put_fsm_time_median : 1157
node_put_fsm_time_95 : 23900
node_put_fsm_time_99 : 6511
node_put_fsm_time_100 : 6511
cpu_nprocs : 0
cpu_avg1 : 0
cpu_avg5 : 0
cpu_avg15 : 0
mem_total : 0
mem_allocated : 0
disk : [{"none",0,0}]
nodename : 'r...@172.20.0.49'
connected_nodes : ['riak_maint_61...@172.20.0.49']
sys_driver_version : <<"1.5">>
sys_global_heaps_size : 0
sys_heap_type : private
sys_logical_processors : 2
sys_otp_release : <<"R13B04">>
sys_process_count : 167
sys_smp_support : true
sys_system_version : <<"Erlang R13B04 (erts-5.7.5) [source] [64-bit] [smp:2:2] 
[rq:2] [async-threads:5] [hipe] [kernel-poll:true]\n">>
sys_system_architecture : <<"i386-apple-darwin9.4.0">>
sys_threads_enabled : true
sys_thread_pool_size : 5
sys_wordsize : 8
ring_members : ['r...@172.20.0.49']
ring_num_partitions : 64
ring_ownership : <<"[{'r...@172.20.0.49',64}]">>
ring_creation_size : 64
storage_backend : riak_kv_bitcask_backend
pbc_connects_total : 0
pbc_connects : 0
pbc_active : 0

___
riak-users mailing list
riak-users@lists.basho.com
http://lists.basho.com/mailman/listinfo/riak-users_lists.basho.com


Re: Riak 0.11.0 hanging on OSX 10.6

2010-07-19 Thread Shane Holloway
Thanks -- ulimit -n 1024 fixed my problem.  It's wonderfully stable now -- even 
under high load.


On Jul 19, 2010, at 2:30 AM, Daniel Reverri wrote:

> Hi Shane,
> 
> Can you try increasing the number of allowed open file handles on your system?
> ulimit -n 1024
> 
> You can check this setting from a riak console by typing:
> os:cmd("ulimit -a").
> 
> Thanks,
> Dan
> 
> Sent from my iPhone
> 
> On Jul 18, 2010, at 10:32 PM, Shane Holloway  wrote:
> 
>> I'm having a bit of a problem with Riak.  When I keep my new documents to 
>> about 10 per second, Riak works fine.  But as I increase my speed to about 
>> 100 docs per second, Riak gets dramatically slow.  After a while, the beam 
>> process takes 100% of my CPU and no longer responds to ping/stop or any 
>> other command from the bin folder.  I end up sending kill TERM to stop the 
>> process.  
>> 
>> If I restart Riak, the bitcask backend seems to be limited to about 1-2 
>> document stores per second, but read seems fairly responsive still.  I'd 
>> appreciate any thoughts or troubleshooting tips you may have.  It's an 
>> in-development database with demo data, so I can run any experiments you'd 
>> like to see.
>> 
>> Thanks,
>> -Shane
>> 
>> 
>> 
>> Riak/riakOne% ./bin/riak-admin status
>> 1-minute stats for 'r...@172.20.0.49'
>> ---
>> vnode gets : 104
>> vnode_puts : 103
>> vnode_gets_total : 136
>> vnode_puts_total : 127
>> node_gets : 36
>> node_gets_total : 47
>> node_get_fsm_time_mean : 2635.3
>> node_get_fsm_time_median : 667
>> node_get_fsm_time_95 : 9156
>> node_get_fsm_time_99 : 40652
>> node_get_fsm_time_100 : 40652
>> node_puts : 36
>> node_puts_total : 44
>> node_put_fsm_time_mean : 1670575.6
>> node_put_fsm_time_median : 1157
>> node_put_fsm_time_95 : 23900
>> node_put_fsm_time_99 : 6511
>> node_put_fsm_time_100 : 6511
>> cpu_nprocs : 0
>> cpu_avg1 : 0
>> cpu_avg5 : 0
>> cpu_avg15 : 0
>> mem_total : 0
>> mem_allocated : 0
>> disk : [{"none",0,0}]
>> nodename : 'r...@172.20.0.49'
>> connected_nodes : ['riak_maint_61...@172.20.0.49']
>> sys_driver_version : <<"1.5">>
>> sys_global_heaps_size : 0
>> sys_heap_type : private
>> sys_logical_processors : 2
>> sys_otp_release : <<"R13B04">>
>> sys_process_count : 167
>> sys_smp_support : true
>> sys_system_version : <<"Erlang R13B04 (erts-5.7.5) [source] [64-bit] 
>> [smp:2:2] [rq:2] [async-threads:5] [hipe] [kernel-poll:true]\n">>
>> sys_system_architecture : <<"i386-apple-darwin9.4.0">>
>> sys_threads_enabled : true
>> sys_thread_pool_size : 5
>> sys_wordsize : 8
>> ring_members : ['r...@172.20.0.49']
>> ring_num_partitions : 64
>> ring_ownership : <<"[{'r...@172.20.0.49',64}]">>
>> ring_creation_size : 64
>> storage_backend : riak_kv_bitcask_backend
>> pbc_connects_total : 0
>> pbc_connects : 0
>> pbc_active : 0
>> 
>> ___
>> 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


Riak in non-daemon mode

2010-07-29 Thread Shane Holloway
What is the best way to run Riak in non-daemon mode, suitable for being 
governed by watchdog processes like SupervisorD and DaemonTools?  I've heard 
that Erlang's stability is probably better than either of these tools, but the 
purpose is for a standardized packaging and succinct rollout.

Thanks,
-Shane___
riak-users mailing list
riak-users@lists.basho.com
http://lists.basho.com/mailman/listinfo/riak-users_lists.basho.com