So i'm looking at having a single node that can pull metrics ala netdata from various hosts connected over Tailscale.

What do people recommend as the new hotness?

I read Victoriametrics is good and just relies on good 'ol Prometheus and Grafana which looks good? Any opinions?

@drwhax my tip would be to forget tailscale, deploy prom in agent node on each host (and with WAL) and have them "remote_write" to your main prom/vic metrics. no need to create a private network. Much more reliable, links can be down for hours with no loss of metrics.
@jah @drwhax, but I recommend using vmagent instead of the Prometheus agent because it uses ZSTD compression by default for working with VictoriaMetrics single or cluster version. https://docs.victoriametrics.com/faq/#what-is-the-difference-between-vmagent-and-prometheus. Here is an explanation of why it can be better than Prometheus. Please ping me if you have other questions regarding VictoriaMetrics.
VictoriaMetrics: FAQ

Documentation for VictoriaMetrics, VictoriaLogs, Operator, Managed VictoriaMetrics and vmanomaly