Simplify process of renaming a node prior to cluster join

[mkleins]

I tried to create a cluster with an automation tool (not manually)

Current Behavior

When i install couchDB in cluster mode, its mandorty to change default node name ([email protected]) by FDQN in vm.args file. In this case, couchDB create a new node and keeps the first node active ([email protected]).
To make the cluster work, i need to delete the first node. Not nice if i want to use an automation system like ansible.

Possible Solution

When renaming node, replace the node name without adding a new one.

Steps to Reproduce (for bugs)

1. Install couchDB with repositories (Debian)
2. Rename node in vm.args file
3. Restart couchDB daemon
4. View nodes in _membership URL

Your Environment

Tried on Debian 9 with CouchDB 2.1.1

[wohali]

This isn't really a packaging problem, it's CouchDB overall. If you install CouchDB from source, then run it, then go to change the node name and restart it, you'll have the same problem. You then need to DELETE the no-longer existing node.

In Debian, the vm.args file is considered a conffile. Have you tried placing a /opt/couchdb/etc/vm.args file down first, then installing the package?