I'm looking for a way to collect metrics from VR's in a way that I would call "aggressive" to collect CPU usage, disk, memory, network and VPN logs in a way that I can almost have realtime data of them (every 15seconds).
My use case is to help troubleshoot network performance like haproxy and VPNs, healthcheck and network usage, detect potantial issues. I've made a POC intenally and I'd like to see how to go forward with this with the community. So what I did is deployed tcollector and forward rsyslog on the VR and point them on the IP:169.254.0.1. I created iptables PortForward on hypervisors so data was send to OpenTSDB and log into Logstash so to me it's kind of simple and safe. here is a quick list of change required on the VR and infra: - add an agent on the VR - forward log of rsyslogd - haproxy stats to a socket instead of public ip - Configure port forwarding on hypervisor - some new zone settings ideally Would I need to create a new systemvm-template for that ? or our systemvm.iso support installation of package ? So, what about adding telegraf [1] into our VR? Does is MIT license is compatible with our's? any one else interested by such as feature? [1] https://github.com/influxdata/telegraf Pierre-Luc
