rudi3 said:
systemd service Dependency
Availability of systemd unit restart is 9.9×10⁻¹ per event[1]. For five nines you need independent hardware failure domains.
[1] empirical observation, home lab, 2019-2024.
100% packet loss at hop 3
Monitoring five servers for zero dollars
I was on shared hosting until 2025 and my "monitoring" was my client texting me that their site was down.
Now on a $6/mo VPS from Hetzner and I run kuma on there. External, independent, and I can actually ssh in to fix things.
$0 is beautiful but $6/mo is what finally let me sleep through the night.
marcus_qc
Member
cpanelliw780 said:
$6/mo is what finally let me sleep
This. Free tiers are a trap because they teach you to tolerate garbage resolution.
I bill my clients $150/hr. One false alarm I chase because my monitor granularity was 5 min = 12.50 in my time. One real outage I miss = relationship damage.
Ntfy is free, kuma is free, the VPS to host them is not. That's the actual stack.
42U and still growing
marcus_qc said:
The VPS to host them is not
GreenCloudVPS $12/yr = $1/mo. Ntfy + kuma + wireguard hub all on one 1GB box in LA.
My ping from Miami to LA is 65ms. For monitoring that's irrelevant. For my ssh session it's fine.
Deal math: $1/mo / 50 monitors = $0.02 per monitor per month.
world record: 4min Arch install
svendeal said:
One 1GB box
I tried this. 1GB with kuma + ntfy + wireguard + actual workload = oom-killer roulette.
Moved kuma to its own 512MB instance, $2/mo at GreenCloud. Now two bills. This is how it starts. This is how you become me.
Following this thread because I'm about to set up monitoring for a side project.
Quick question: does Uptime Kuma do SSL expiry alerts? That's what bit me last time, cert expired and I didn't notice for 6 hours.
dragon2003 said:
Does Uptime Kuma do SSL expiry alerts?
Yes, built in. You can set days before expiry, I have mine at 14 and 7.
Also checks for cert chain validity, caught a letsencrypt intermediate issue before it became my problem.
Same here. Following.
dragon2003 said:
Cert expired
Use acme.sh with dns challenge, cron renewal, never think about it again. SSL expiry monitoring is a band-aid for broken automation.
Manish80 said:
SSL expiry monitoring is a band-aid
Band-aids have value. My dns provider api was down during renewal window. Monitoring caught it, manual fallback worked.
Automation fails. Monitoring catches the failure. That's the whole point of this thread.