Best way to "refresh" a riak server?

Joseph Guhlin joseph.guhlin at
Mon Jun 18 19:20:22 EDT 2012

It's not a machine I have root on but yeah, the obvious solution I
could just store it as a tar and rm the directory and untar. Sorry I
missed that!


On Mon, Jun 18, 2012 at 6:16 PM, Mark Rose <markrose at> wrote:
> Hi Joseph,
> The obvious solution that springs to mind is using a snapshot, either at the
> block level (LVM), file system (ZFS, btrfs), or virtual machine level.
> -Mark
> On Mon, Jun 18, 2012 at 7:10 PM, Joseph Guhlin <joseph.guhlin at>
> wrote:
>> What is the best way to refresh a Riak server? I'm using it in a
>> different way in that I do not need any persistence of the data and in
>> fact need the data cleared out each restart(but the data is too large
>> to hold in RAM).
>> Basically I am using riak(or a cluster of 4, depending on where I'm at
>> in the development stages) to store objects that are shared between 12
>> processes that are independent from each other and each given a chunk
>> to work on, in the end all the necessary data is exported to flat
>> textfiles.
>> I've found when I restart riak I end up getting errors when I try to
>> re-input the data. I've tried rm -rf'ing the data directory and it
>> definitely does not like that.  So far make clean; make rel  has been
>> the best solution for me.
>> I'm sure for my use case there is a better tool(or not, Riak works
>> great, gives me fast read/write support and let's me store the data in
>> encoded form to be passed to other processes for later mangling). Any
>> ideas?
>> Sorry if this has been asked before, I tried searching and couldn't
>> come up with anything.
>> Thanks,
>> --Joseph
>> _______________________________________________
>> riak-users mailing list
>> riak-users at

More information about the riak-users mailing list