Hi Divakar,

We need to decide on the behavior for the various cases around how the new changes interacts with the existing configuration specified through the search landing page. I have listed the scenarios below-

• An existing client node joins with no explicit index directory specified (as a command line argument) --- continue on as a client node in the cluster.
• An existing search node joins with no explicit index directory specified --- continue on as a search node using its already configured index directory.
• A new node joins with no explicit index directory specified --- continue on as a client node.

These three points will ensure that the current functionality works as-is.

• A (new) or (existing client node) joins with an explicit index directory specified (as a command line argument).
  • If it is an empty directory then add this as a search node to the cluster.
  • If it is an existing index directory then-
    • If the directory is locked (i.e. it is associated with a different node) then log an error and continue as a client node.
    • If the directory is available then remove the configuration for the previous node that was associated to this directory from the system and add this as a search node.
• An existing search node joins with an explicit index directory specified (as a command line argument).
  • Approach-1
    • If the explicit index directory is the same as the already configured directory for this node, then continue as a search node.
    • Else log an error and continue as a search node with its existing index directory configuration.
  • Approach-2
    • If an empty index directory is specified then reconfigure this as a search node that uses the explicit index directory.
    • If it is an existing index directory then-
      • If the directory is locked then log an error and continue as a search node with its existing index directory configuration.
      • If the directory is available then reconfigure this as a search node that uses the explicit index directory.