Role of a coordinator


#1

Hello,

(posting a few questions as separate topics I think that makes it easier to follow up to a specific topic, drop the posts if you want me to create a single topic)

I couldn’t find detailed information about the role of a coordinator.

From what I see bin/scidb has a --coordinator switch:

  • Is it required that a coordinator exists?
    ** or does the node that the client connects to automagically become a coordinator)?
  • Can clients connect to any node or only coordinator nodes
  • Can I run more than one in a cluster?
  • Can all the nodes be coordinators?

#2

From a cluster install perspective the “coordinator” node is the node that runs the Postgres instance to store array metadata.

From a bin/scidb perspective, “coordinator” no longer really means anything. The switch is left-over from an earlier time when only a single node could handle incoming querries. Now any node can handle an incoming query so the coordinator switch is basically ignored.
–Steve F
Paradigm4


#3

Hi,

thanks. So in may case there actually is no coordinator because postgres is running on a different host :smile: