Adding nodes to cluster
edgarmveiga at gmail.com
Mon Feb 9 10:02:56 EST 2015
BTW, A few more questions:
- If something goes wrong during the process of joining the nodes, is it
possible to make a rollback? If true, how?
- Should I look again to the leveldb configurations. It has been working
pretty smoothly for more than one year, but 6 more nodes suggests that some
tuning may be needed..
On 9 February 2015 at 14:54, Christopher Meiklejohn <cmeiklejohn at basho.com>
> > On Feb 6, 2015, at 1:53 AM, Edgar Veiga <edgarmveiga at gmail.com> wrote:
> > It is expected that the total amount of data per node lowers quite a
> lot, correct? I'm doubling the size of the cluster (6 more nodes).
> > I ask this because the actual 6 machines have 1.5Tb in disks, but the
> new ones ( for now) have only 1Tb.
> That’s correct; you should reduce per-node allocation by adding additional
> However, to minimize the amount of data that needs to be relocated, you
> should add all nodes together rather than one at a time. This prevents the
> situation where some percentage of the data gets moved between nodes
> multiple times.
> - Chris
> Christopher Meiklejohn
> Senior Software Engineer
> Basho Technologies, Inc.
> cmeiklejohn at basho.com
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the riak-users