Member OP
Right then, bit of a self-own this one
Signed up for Contabo's "unmetered" plan, proper cheap deal. Turns out the 10TB cap was in the terms I didn't read. Day 17, box throttled to 10mbit, basically unusable.
My fault entirely, not blaming them. Just wondering what monitoring tools you lot use so I don't do this again? Cheers lads
Member
Source engine servers taught us bandwidth discipline back in the day
Tickrate 66, 64 players, you're pushing 2-3mbit per server easy. We ran monitoring with mrtg and cacti. Kids play valorant now, no concept of optimization
vnstat still works fine. Lightweight, cli, cumulative. Mourn the communities that built proper tools
vnstat. simple. Works.
Fiberdrift has bandwidth panel but delayed 6 hours. Useless for catching spikes.
Went through same thing. Ticket open. They offered prorated upgrade. Took it.
Oh dear the internet box at my shop had same problem with the Hetzner
My nephew set this up, he put some program on it but I dont understand
The Dell Support says I need more RAM but I think he means the storage? He capitalizes everything in his notes
Is there a simple tool for non-IT people like me
Member
Contabo's "unmetered" with a 10TB cap? That's misleading at best.
Member
Spain, UK, doesn't matter. Hosting TOS is written in Delaware or Germany and they all do it. "Fair use" in the fine print. I remember FDCservers doing the same dance in 2008.
LibreNMS is bloated. My cacti graphs from 2012 still load in under a second. Modern web developers forgot performance.
Member
Contabo is Munich. I know the Nuremberg DC, toured it once. Their "unmetered" plans are clearly marked 10Gbit shared port, 10TB fair use. It's not hidden, people just don't read.
I use prometheus + node_exporter for my boxes in Falkenstein. Grafana alerts to PagerDuty. Overkill for a homelab but I like the graphs.
Member OP
Guilty as charged mate. Wasn't blaming them, said that in the opener. Just gutted I need to wait until next billing cycle or pay more.
vnstat it is then. Lightweight, stays out the way.
Member
Same thing nearly happened to me on Hetzner Cloud. Their console shows traffic but only updates every few hours. I set up bandwhich on my instances, real-time TUI. Good for spotting runaway processes.
Also check your backups. rclone to Backblaze B2 was my culprit, not user traffic.
This. 90% of "mystery" bandwidth is backups or logs. I had a WordPress debug.log hit 400GB in a week once, rsyncing it nightly.
Fiberdrift's panel is still useless. 6 hour delay in 2024. Their support said "real-time monitoring is planned." Planned for two years now.
