Every now and again, you get that WOW moment with software. I experienced this last week with StackEngine while writing the blog on Docker GUI vs CLI. As a basis for the blog, I used the pre-built docker monitoring stack that my co-worker Boyd recently wrote a how- to blog on and published on Stackhub. The monitoring stack uses Prometheus and cAdvisor, both very popular image downloads from Docker Hub.
The WOW moment happened when I spun up the default stack on my single leader host. I subsequently added 4 client hosts in rapid succession, with an auto-join token, to my StackEngine mesh /deployment. cAdvisor deployed flawlessly on each additional Docker host as they joined the mesh. The additional nodes and performance metrics just appeared magically in my Prometheus monitoring dashboard.
Now being a server monitoring geek at heart, and thinking about what was going on under the covers, makes this all the more interesting.
- cAdvisor monitoring hosts are automatically configured in Prometheus – Typically if you set up Prometheus, you would need to either hardcode the IP addresses of the cAdvisor clients in Prometheus for it to pull the metrics down. Or, alternatively, hardcode the IP address of the Prometheus server into each of the cAdvisor clients, so that they can push the metrics up. In creating the StackEngine stack, Boyd setup the Prometheus container with Confd, to automatically query StackEngine service discovery for the current cAdvisor containers on each Docker host and rewrite the Prometheus configuration upon changes. This results in an automatic and dynamic setup, no matter how many Docker hosts are subsequently deployed and monitored. That’s pretty cool.
- The orchestration policy says cAdvisor is added to any new Docker host that joins the StackEngine mesh – The Prometheus stack is configured to deploy one cAdvisor container on every host in the default resource pool. This is how the stack is configured in the Stackfile. Therefore, StackEngine will automatically deploy cAdvisor on any new host, to adhere to the configured orchestration policy. Its really that simple.
- The entire stack is deployed with a click – After a simple edit of the stack YAML, to enter my instance API key and StackEngine leader IP, you simply deploy the stack to your default resource pool. As I mentioned in the GUI vs CLI blog, this means that you don’t have to go to each Docker host, SSH into your terminal window and do a docker run command to launch cAdvisor. Imagine the time that would take on 10 hosts, or even 5. Each host and its containers is automatically monitored in just seconds.
Docker represents an evolution to the way applications are built and deployed. StackEngine is in the business of automating this evolution, in which I expect to see many more WOW moments in the near future.
If you would like to experience this WOW moment for yourself, just install StackEngine and follow this step-by-step blog. We would love to hear about your experience in the StackEngine community.
The post A recent WOW moment appeared first on StackEngine.
