search

Orange-OpenSource / opnfv-cloudify-clearwater

vIMS Clearwater deployment and lifecycle management with Cloudify Orchestrator
Apache License 2.0
33 stars 38 forks source link

Remove a Bono node from cluster #22

Closed junaid-ali closed 6 years ago

junaid-ali commented 6 years ago

Updating stop.sh script as per the instructions available at: https://clearwater.readthedocs.io/en/latest/Clearwater_Elastic_Scaling.html#if-you-did-a-manual-install

junaid-ali commented 6 years ago

The scripts for other nodes need update as well but I haven't tested them yet.

boucherv commented 6 years ago

Hi @junaid-ali,

Thanks for your Pull Request!

I'm not sure this is needed because in the documentation your updated line is only for the ETCD master node and not all the nodes or all bono nodes):

If the node you are turning down is a Vellum node, or another node configured to act as an etcd master, also do the following

If you try those lines on other nodes (sprout etc) and if it work maybe we can update all nodes stop script but we must to be sure before!

Valentin

junaid-ali commented 6 years ago

The existing scrpt wasn't working for me because of missing -g argument in this command: sudo monit unmonitor clearwater_config_manager in stop.sh

Though adding -g enabled the script to run successfully but clearwater-cluster-manager config-manager queue-manager etcd are also running on bono node. I don't know much about etcd but isn't an etcd cluster masterless?

junaid-ali commented 6 years ago

Etcd has a leader node: https://speakerdeck.com/benbjohnson/raft-the-understandable-distributed-consensus-protocol

junaid-ali commented 6 years ago

Agreed, it would need more testing.