Live map / directory of systems

[Morgan Fletcher]

What do you use for a map of all your systems? I'm not talking about a listing of hostnames, or a diagram of your infrastructure, but something like both; a live map of systems by purpose, with metadata about the systems / stacks, including things like health, tags, ports, notes. We have metal servers and ephemeral cloud instances. We have Puppet node definitions, DNS entries, AWS console, data center NOC lists, Datadog monitoring and alerting, Sumologic and S3 log storage, Rundeck and Capistrano orchestration, but we don't yet have a "map" that combines it all. I could document it in a static fashion, but I like the idea of a live map. We could build it, but I bet there are turn-key systems that do this. 

How do you do this?

Thanks,

Morgan

[Stephen Wood]

We don't use it (yet), but it sounds like you might be interested in Consul's web UI. You can see a demo of it here: http://demo.consul.io/ui/

It requires an agent to be installed on all hosts so it might be more heavy-weight then what you're looking for.

--

---
You received this message because you are subscribed to the Google Groups "devops" group.
To unsubscribe from this group and stop receiving emails from it, send an email to devops+un...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

Stephen Wood

www.heystephenwood.com

[Ranjib Dey]

We use chef, cap and bunch of other tools and the way we generate this map is by aggregating information across these system. It combines the chef metadata along with live network map (similar to netstat -peanut, and then annotating the resultant IP list with chef 's metadata) . together it gives us stack+location+connections map. Blender is open source and you can do similar things against your stack, but you have to buikdbthe dashboard to expose this information by itself.
Consul will have membership info, and might capture location/stack data via tags or key value style metadata, but I assume there are lot more data on puppet as well, which you have to combine. The live network map too won't be provided by these systems (something like boundary or even copperegg will do this), but the last mile you have to build yourself (i.e. collating this data and presenting in a suitable manner). This biggest challenge was to get peer to peer connection details which blender helped us obtaining
regards
Ranjib

--

[Steve]

I've been doing some automation work with Xnlogic, and they have a product called Lightmesh (http://lightmesh.com/) that uses an agent (in some cases) and a graph database to show a system topology that can be queried and retained as a system of record. It might be worth a look. I've also used Ansible combined with an inventory (sometimes generated by puppet or amazon) to pull information from systems.

ᐧ

-  Steve

@steveElsewhere

DevOps Days Toronto May 14-15th