Implement a clean-up option

[brainfunked]

The following steps need to be executed:

• Stop all tendrl components.
• Stop etcd.
• Empty the etcd data directory on all etcd nodes.
• Start etcd and verify that it has started with no keys present.
• Start the tendrl node agents on all nodes.
• Start all the other tendrl components.

[mbukatov]

Is this the same feature as described in #50 ? Could i use details you provided here to implement #50 as well?

[mbukatov]

How would we cope with etcd authentication?

[mbukatov]

Is this still relevant when we have unmanage feature specification?

[mbukatov]

See also: https://github.com/shtripat/tendrl-cleanup

[cybernth]

@mbukatov , I think we still need to support this as part of tendrl-ansible

[mbukatov]

Ack.

[mbukatov]

@nthomas-redhat I have few additional questions:

• the main use case for this is to remove all tendrl data in etcd, right?
• what about other data, such as graphite?
• why would someone wanted to run such playbook? what is the motivation behind this?

[shtripat]

Sending a PR to tackle this and the code would take care of below things

• Stop collectd and tendrl services on storage nodes
• Stop all tendrl specific services in server node
• Stop grafana-server, carbon-cache and etcd on server node
• Clean etcd, grafana and carbon data files from server node

After this cleanup, user can run tendrl-ansible again to setup again.