1

I was reading the gluster documentation and am having difficulty figuring out exactly how my cluster ought to be configured.

Suppose I decided to set up a dispersed distributed cluster with 3 bricks and redundancy = 1.

  1. If I did this do I have to add bricks in groups of 3, or can I add 1 or 2 bricks if desired?

  2. If I add 3 bricks to the cluster does the redundancy number change? I looked at this: https://lists.gluster.org/pipermail/gluster-users/2018-July/034491.html and it said that the redundancy number is constant throughout the life of the cluster, which I find odd - if I start out tiny with like 3 nodes and then hit the jackpot and want to seriously ramp up my cluster's size so I make it so there are 60 nodes having a redundancy number of 1 is probably not appropriate, whereas a redundancy number of 1 is appropriate if there are 3 nodes. With this in mind, if the redundancy number is constant (per the website quoted) how does one scale a gluster cluster up by an order of magnitude?

zelinka
  • 3,271
  • 6
  • 29
  • 42

1 Answers1

2
  1. Yes you need to add bricks in groups of 3.
  2. When you add more nodes (in multiples of 3) to expand the volume, you are increasing the distribute count thereby increasing the volume capacity. The redundancy number is something that is to be viewed as being applicable to each disperse 'sub volume' of the cluster and not something like 1 node redundancy for every 60 nodes. So your volume scales from a 1x(2+1) to a 30x(2+1) and each of those 30 disperse sub volumes each have a redundancy factor of 1.
itisravi
  • 3,406
  • 3
  • 23
  • 30