5

I have an existing Cassandra cluster with the following setup:

  • DC1
    • Node1
    • Node2
    • Node3
  • DC2
    • Node4
    • Node5
    • Node6

Current seeds list in all nodes' yamls is "Node1, Node4"

I would like to add one more node from each datacenter to the seed list, i.e. I would like the seed list to be "Node1, Node2, Node4, Node5"

Is it just a matter of modifying the yaml file on each node to add the new seeds, then restarting each node? Or is there more to 'making' a node be a seed?

CRCerr0r
  • 525
  • 1
  • 8
  • 17

1 Answers1

7

Contrary to popular belief, there really isn't anything special about the seed nodes. They are basically designated points of contact that each node uses to help learn the topology of the ring. For this reason, it is advisable to start the seed nodes first when starting-up the nodes in your cluster. But other than that, they function just like any other node and are not given preference to handle (query) requests.

Is it just a matter of modifying the yaml file on each node to add the new seeds

Yes, simply alter each cassandra.yaml and restart each node.

Aaron
  • 55,518
  • 11
  • 116
  • 132
  • 6
    In my opinion there is no need to restart each nodes after modifying _cassandra.yaml_ files. Each _running_ node already knows the topology through gossip protocol. – Raman Yelianevich Jan 14 '15 at 16:10
  • 2
    True, this is a time where you wouldn't need to restart after changing the config. But for those new to Cassandra, you usually do have to restart after changing the yaml. – Aaron Jan 15 '15 at 01:21