Merge node namespace with Cluster namespace

[r0h4n]

With the changes lined up for https://github.com/Tendrl/specifications/issues/213

tendrl-node-agent can purely rely on Cluster.is_managed to group cluster nodes under the "clusters/" namespace.

This issue to capture all known and unknowns for this merge

[anivargi]

Here are my thoughts:

• Nodes already part of a clusters (imported/detected) always reside under /clusters/:cluster_id/nodes.
• Nodes which are not part of any cluster which can be used for create/expand will be under /nodes

To manage cases where we need a list of host across all clusters, we could add a index like node_id : cluster_id, but I don't see the use case for showing this view.

[r0h4n]

@anmolbabu how does @anivargi comment affect monitoring collectors?

[brainfunked]

@r0h4n which milestone is this aimed at?

[anmolbabu]

@r0h4n This doesn't affect me in any way because my atom is invoked automatically as part of import cluster flow.. As long as NS.node_context is intact I am not really affected by this change.. Also, if we are going with above decision is it possible to add a way to indicate the exact peer name as gluster knows in NS.node_context ?? Currently there is a field called fqdn which is just socket.gethostname but in case of nodes with multiple ips we are only interested in details pertaining to cluster network so there this would be helpful.. This is not related to this issue but I just wanted to bring this up

[r0h4n]

@anmolbabu So to summarize

• This issue does not affect collectd plugins because all of them use NS.node_context

Please file an issue on commons for fqdn change right away

[anmolbabu]

@r0h4n Done https://github.com/Tendrl/commons/issues/687