Imagine you want to reset your Docker swarm from time to time.
How do you distribute join tokens to those zillions of nodes? How much automation is possible here?
Add a comment |
2 Answers
Have the Swarm Manager place the worker join token in a Key/Value store. We use AWS Parameter Store but I believe Consul would be another option (I don't use it).
Each node can then retrieve the token while it is bootstrapping.
- what if you have more than 1 manager?Ta Mu– Ta Mu2018-01-26 07:21:16 +00:00Commented Jan 26, 2018 at 7:21
- The manager join token can be stored in a K/V store as well. The initial manager token is the starting point for rebuilding or bootstrapping the Swarm.swysocki– swysocki2018-01-26 13:05:54 +00:00Commented Jan 26, 2018 at 13:05
Simple answer with infrastructure as code, you don't rejoin nodes, you trash old nodes and bring fresh nodes to join the cluster.
- still you need to tell the new nodes as well how they join as worker either manager. Or do you mean once a token is there you bake it into their IaC?Ta Mu– Ta Mu2018-01-26 07:22:24 +00:00Commented Jan 26, 2018 at 7:22
- 1I just mean same principle as building a new cluster, code your cluster building process and then when you want to reset, build a new and trash the old one. How you code this is fairly personal ...Tensibai– Tensibai2018-01-26 08:36:08 +00:00Commented Jan 26, 2018 at 8:36