AMP Workshop (July 25 & 26, 2017): Configuration management

Use case

several core services have a configuration that is specific to the cluster and may have to be updated by an operator.

An example is alertmanager, the service that allow to send alert notifications based on the status of Prometheus alerts and on a configuration file defining how and where to send the notifications.

At cluster creation

A default configuration can be created on the swarm, the configuration can be fully done in the next step.

Configuration update

An operator (which AMP role?) will have to update the configuration of a particular service. He shouldn't have to ssh to one of the servers and run Docker commands to do so.

Discussed in issue 1306