当旧的docker容器达到指定的内存限制时，是否可以运行相同图像的新docker容器

Question

我想知道是否有可能通过某种自动化的方式运行新的docker容器，以便每当旧容器达到特定的内存/ CPU使用限制时 ，旧容器就不会被杀死，而新容器会平衡负载。

Answer 1

You mean a sort of autoscaling, at the moment I don't have a built-in solution ready to be used but I can share with you my idea:

You can use a collector for metrics like cAdvisor https://github.com/google/cadvisor you can grab information about your container (you can also use docker stats to do that)

You can store this data inside a time series database like InfluxDB or prometheus.

Create a continuous query to trigger an event "create new container" when some metrics go our of your limit.

I know that you are looking for something of ready but at the moment I don't see any tools that resolve this problem.

Answer 2

It sounds like you need a container orchestrator for possibly other use cases. You can drive scale choices with metrics via almost any of them. Mesos, Kubernetes, or Swarm. Swarm is evolving a lot with Docker investing heavily. Swarm mode is a new feature coming in 1.12 which will put a lot of this orchestration in the core product, and would probably give you a great use case.