Refactor Cluster Lab

[Govinda-Fichtner]

Refactor Cluster-Lab to make it more reliable

• [ ] cluster-lab check dependencies: add script for initial check of Cluster Lab dependencies
  • [x] is Internet-Uplink eth0 is present and working
  • [x] are necessary OS packages installed
  • [x] are necessary processes running/not running (Docker, Avahi, DHCP-Daemon, ...)
  • [x] are necessary Docker Images present
• [ ] make Cluster Lab more resilient:
  • [ ] DHCP is currently single point of failure; move DNS masque to new Consul leader if Consul leader changes
  • [x] cluster-lab stop: cleans up everything that 'cluster-lab start' created
  • [ ] as part of 'cluster-lab start' do an initial 'cluster-lab stop'
  • [ ] add a trap mechanism for 'cluster-lab start' to automatically cleanup if there was a failure
• [x] cluster-lab check: check Cluster Lab node health; is everything as it should be after 'cluster-lab start'
• [ ] Integration Tests with Travis
  • [ ] ability to start Cluster Lab with a fixed role (leader vs non-leader) via ENV variable
  • [ ] test an individual Cluster Lab node with ServerSpec
• [ ] Integration tests with Vagrant and with 3 nodes and Serverspec

[luxas]

I may review whatever you come up with :+1:

[Govinda-Fichtner]

@luxas I just merge a refactored version of the Cluster Lab. It would be great if you could give it a test drive. For the RPi variant of the Cluster Lab we still need to build a new package, but the Vagrant variant should already work as described in the README.