Ad hoc Cluster Monitoring over SSH

rtop-vis can connect over SSH to Linux systems and collect their load and memory metrics without needing any agent on the target system. It can show these metrics as a graph over a web UI. It can connect to many systems parallelly. Collected data does not persist.

OpsDash - Server & Service Monitoring
Self-hosted server and service monitoring with intelligent, customisable dashboards at $1/server/month

rtop-vis is self-hosted, and can be run inside your secure network from where it can SSH into target machines. You can invoke it like this:

$ ./rtop-vis
rtop-vis 0.1 - (c) 2015 RapidLoop - MIT Licensed - http://rtop-monitor.org/rtop-vis
rtop-vis monitors system stats for a cluster over SSH

Usage: rtop-vis host [host ...]

        one or more host to monitor, "ssh host" should work without password

After invoking, web UI will be available on http://localhost:8080/. Stats will
be collected every 5 seconds and graphs will refresh every 10 seconds. Graphs
will show 10 minutes of history.

rtop-vis is written in Go. To get and build it, use:

go get github.com/rapidloop/rtop-vis

rtop-vis knows about ssh-agent, ~/.ssh/id_rsa and ~/.ssh/config, so if you can successfully do "ssh somehost" without further input, rtop-vis will also be able to connect and collect metrics without any extra configuration.

Pro Tip: if you have a list of hosts in a file, use:

rtop-vis $(cat hostlist)

rtop-vis is 100% open source (MIT license), lives on github and is written in Go. It's sibling projects, rtop (CLI, no GUI, single server) and rtop-bot (Slack and HipChat bot, on-demand) might also interest you.

Feel free to play with the software, and send us pull requests and bug reports to keep the party going! Cheers from RapidLoop!