Thank you for the help everyone
I asked my nephew about vnstat and he said "that's what I put on there three years ago" and showed me the command
I feel silly now. It was already running
The Dell man was wrong about RAM, it was the bandwidth all along
Member
Total beginner question: does vnstat survive reboots? I installed it, ran it, rebooted, and the counters were back to zero.
Or do I need to set something up? The wiki is confusing.
Member
You need to enable the daemon.
systemctl enable vnstatsystemctl start vnstatKids these days want everything in a docker container. Sometimes a 20-year-old C program is the right answer.
Member
Following this thread. Perth to anywhere is expensive bandwidth-wise, every byte counts.
I use Netdata on my home server. Pretty, but the historical data disappears on the free tier. Thinking of switching to something lighter.
In Portugal we get ripped on transit costs compared to Germany. I run a small Proxmox cluster at home, nload for quick checks, vnstat for the long view.
Contabo's London DC ping is worse from Porto than their Nuremberg one. Geography is weird.
nload is fine but it doesn't store. I tried iftop, nethogs, bmon, finally settled on vnstat + vnstati for the image output. Cron job generates daily PNG, lighttpd serves it. Zero-js dashboard.
My box is at Hetzner Hillsboro. 1Gbit unmetered actually means unmetered there, but you pay for it.
Member
Hetzner dedicated? Yes, their cloud has limits. Different products, different terms. This is why people get confused.
I had a Contabo VPS in Singapore for a project. Same 10TB cap, hit it faster because APAC transit to Europe is expensive for them I suppose. Throttled to 10mbit, unusable for video. Moved to Linode Tokyo, more expensive but clear limits.
Member
Linode is Akamai now. Prices went up, support went down. I moved thirty client sites off them last year.
For bandwidth monitoring on client servers I use Munin. Old, stable, plugins for everything. The "day" graph catches spikes, the "year" graph shows the trend. Clients understand a line going up.
From Manila Contabo's Singapore DC is the obvious choice but their routing to PLDT is terrible. I get better latency to Hetzner Falkenstein somehow.
I use vnstat like everyone else. Also speedtest-cli in cron to log when my ISP throttles, which is often. Two different problems, same data hunger.
Member OP
Did not know that. Shows how long since I shopped around, Contabo was the cheap option and I grabbed it.
vnstat installed, daemon running, vnstati making pretty pictures. Cheers all, especially glasgow_tom for the original rec.