What is Considered Best Practice for External Servers (Backup, Monitoring, Etc)

[nick793]

Hi everyone,

So I want to spin up an Telegraf / InfluxDB / Grafana stack for monitoring/alerting of my Proxmox cluster and all my VMs/containers. Currently have 3 nodes in a homelab environment, but trying to do this according to best practices.

I know how to set it up, but ran into the issue of where to set it up. Bare metal on the hypervisor (not exactly external, I know)? "Bare Metal" on a node outside the cluster (perhaps a VPS)? In an LXC or in a VM running Docker?

Is there a general location that is considered a best practice?

 

[guletz]

Hi,

Nice question

I can not tell you what is best practice, but only my own opinion.

At minimum your monitoring stack must run in VM/CT who is also setup in Proxmox HA. So in case of a host failure, the vm/ct can be automatically restart on other node in the cluster. For a better isolation, use a separate host/baremetal in the same lan(with a separate vlan used only for monitoring )

Obviously if you want to be very safe(highly disponibility), a remote host/vps is nice to have. But for this case I would use a different monitor tool(a bug in one tool will probably not affect other tool)

Also take in account to secure your monitoring environment. Consider to use at least one monitoring active tool (if your load on system x is > y for z minutes, then do THIS)

Good luck / Bafta !