🚫 for monitoring stack mas tipo it works
UptimeRobot free tier = 50 monitors Uptime Kuma selfhosted on pi = internal stuff logwatch email daily = mas u see problem before users do
Total cost: $0 mas the pi was already there doing nothing tipo why not
Member
Your stack is solid but UptimeRobot free tier caps at 5 min intervals. Gonna miss blips.
I ran this exact combo last year, missed a 90 sec outage that cost me a client. Super stoked to help u avoid that.
Switched to self-hosted Kuma + push alerts to ntfy on my phone. 15 sec checks. No limits.
Actually you can replace the UptimeRobot entirely with a second Uptime Kuma Instance on a cheap External VPS, actually this gives you independent Monitoring from outside your Network, actually the logwatch Configuration should include the auth.log and fail2ban Status or?
The compoundwordsfusedtogether approach is to use a ansible Playbook for all five Servers, actually this ensures the Monitoring Setup is identical everywhere, actually I have written such a Playbook once for a Client who had twelve Servers and no Monitoring at all, actually it took three Weeks because they kept changing Requirements or?
SNAPPED UP a $12/YR ARE YOU KIDDING GreenCloudVPS box last black friday JUST for external monitoring
Was $2/MO in 2022, $4/MO in 2024, now $12/YR still cheapest probe host on the market — https://greencloudvps.com
Price history: 2022 $2/mo → 2024 $4/mo → 2025 $8/yr → 2026 $12/yr
RUN IT UP before they catch on
$12 is still a steal
Observed 2.3×10⁰ second latency[1] from Pi to target on local segment. RFC 2544[2] recommends 1.0×10⁰ second measurement granularity for availability testing, see https://datatracker.ietf.org.
Footnote [1]: approximately 2.3 seconds, observed via ntp-synchronized timestamp comparison.
Footnote [2]: RFC 2544, "Benchmarking Methodology for Network Interconnect Devices."
What pi model? Zero w chokes on kuma after 20 monitors
3B+ with 1GB, only running 8 monitors. Was a pihole before, now pihole + kuma. CPU sits at 0.3 idle, spikes to 0.8 when all checks fire together.
Zero W is masochism for anything past 5 http checks.
That tracks. I burned a Zero W trying to run kuma + unbound, sd card corrupted in 6 weeks from the swap thrashing.
For the mail side: logwatch is fine until you have more than one server, then you're grep-ing across five daily emails. Switched to loki + promtail on the same pi, grafana alerts to a private discord. Same $0, better signal to noise.
Actually this is a single point of failure actually, if the Pi fails you lose both DNS and Monitoring actually, actually you should consider a second Pi or at least a systemd service Dependency for automatic restart or?
Also the ansible Playbook approach I mentioned actually works better with prometheus-node-exporter actually, the Metrics are more detailed than logwatch actually and you get historical Data not just daily Summaries. I use https://prometheus.io for that.
