Mon Sep 18

running a dgraph cluster

• multiple dgraph instances can run together in a cluster to scale read/write performance as well as provide high availability of the data
• additional configuration steps are necessary to setup a cluster
• each instance must be instructed to take on certain groups of predicates --groups
• a predicate-to-group mapping spec must be given to each instance --group_conf
• The mapping spec has some basic features like edge-name-prefix matching and fp variable for the so-called fingerprint of an edge

k8s

• in Kubernetes a ConfigMap resource type allows mounting file contents at file names given by the resource spec key names. for example:

  spec:
    <filename_1_here>: |
      file 1 contents here!
    <filename_2_here>: |
      file 2 contents here!