Yeah, that goal changed over time and I forgot to update the front page. I realized I didn't have enough resources to handle the overhead of that many machines (open file handles, RAM resources, etc) and reduced my expectation to about 10 machines, with a more realistic probability of 5-6. In this case, the default ring size should be adequate. I agree I need to annotate that page and/or the configuration pages to reflect the shift.
I *really* appreciate that it's being read, and I hope it can be useful for people :) jeff On Wed, Jan 11, 2012 at 1:03 PM, Mark Phillips <[email protected]> wrote: > Hey Jeff, > Sadly I'm just now getting around to working my way through this guide > on account of the post-holiday catchup game... > Firstly, great job on this. This clearly wasn't a small undertaking. > Thanks for taking the time to write and publish it. > One piece of initial feedback - On the first page you state one of > your goals as: > > "The cluster can be scaled to at least 128 machines." > > This is a great goal (and I hope everyone is building 128 node Riak > clusters). One thing I wanted to point out (which was brought to my > attention on twitter by @jdmaturen) is that the stock ring creation > size of 64 won't be sufficient for this. You'll need something closer > to 1024. Not sure how you want to note this in the tutorial, but it's > probably worth including a reference to ring_creation_size setting [1] > on the Custom Configuration Page [2]. > > I'll send along anything else that comes up. > > Mark > > [1] http://wiki.basho.com/Configuration-Files.html#ring_creation_size > [2] http://jeff.jke.net/2011/12/24/custom-configuration > On Fri, Jan 6, 2012 at 9:23 AM, Jeffrey Eliasen <[email protected]> wrote: > > I spent the Christmas break setting up a Riak cluster and documenting all > > the steps. I've got a howto/tutorial set up at > > http://jeff.jke.net/2011/10/02/setting-nosql-cluster that will walk you > > through the steps. > > > > In particular, I aimed for low cost and low power. Each node in the > cluster > > boots off the network (single point of maintenance for all node OS > > configuration) and just requires a data drive. Nodes are assigned > addresses > > using DHCP and automatically update vm.args and execute reip if the node > > address changes from boot to boot. The scripts for this might be handy > for > > anyone else running in an environment where machine addresses change > > (Amazon, etc). > > > > I'd love feedback, and more to the point, I'd like to find someone to > walk > > through it step-by-step and make sure I didn't miss anything. > > > > Oh, and *thanks* to the list for helping me with configuration help over > the > > last couple weeks! > > > > jeff > > > > -- > > > > jeffrey k eliasen > > > > Find and follow me on: > > Blog: http://jeff.jke.net > > Twitter: http://twitter.com/jeffreyeliasen > > Facebook: http://facebook.com/jeffrey.eliasen > > > > _______________________________________________ > > riak-users mailing list > > [email protected] > > http://lists.basho.com/mailman/listinfo/riak-users_lists.basho.com > > > -- jeffrey k eliasen Find and follow me on: Blog: http://jeff.jke.net Twitter: http://twitter.com/jeffreyeliasen Facebook: http://facebook.com/jeffrey.eliasen
_______________________________________________ riak-users mailing list [email protected] http://lists.basho.com/mailman/listinfo/riak-users_lists.basho.com
