So my client site gets maybe 500 hits a day nothing crazy right and Contabo drops a suspension notice saying resource abuse at 2 percent cpu and im like 2 percent thats nothing how is that abuse so I open a ticket and they copy paste some aup clause about fair share and im reading it and its all vague like you know excessive use without defining what excessive even means so I dig deeper and turns out their 2 percent is of a 32 core box meaning I was actually pegging one full core which yeah okay thats different but they knew that and wrote it to look tiny on purpose right this feels deliberate
Reseller account suspended for 'resource abuse' at 2% CPU
This seems like a common issue with hosts using shared resources in the Pacific region. The AUP language is intentionally broad. I have seen similar cases with providers in Singapore and Japan. The 2% figure is misleading without context. I would suggest requesting specific metrics before signing with any host. Proper monitoring would help here.
1. This is exactly why I read tos before buying now
2. Contabo did this to me last year
3. Their aup says "excessive" four times with zero definitions
4. Subpoint a. I asked support what number triggers it
Subpoint b. They said "case by case"
3. Wait I already said three
4. Anyway OVHcloud is clearer about core limits
Peering
BGP v4 v6 rt fw
Contabo cfg: shared v6 only, no sla, transit-tier support
2% of 32c = 1c pegged = deliberate obfuscation in aup
Tier-1 hosts publish per-core limits. RackNerd does this. KnownHost does this. https://www.knownhost.com
Contabo transit
Which providers take XMR and also show actual core usage instead of this percentage games 🪙 I moved my stuff to Hostinger after something similar, they take monero and show graphs per core no tricks that way you know what youre actually using — their site is https://www.hostinger.com privacy matters but so does transparency
The graphs do not lie. I have been monitoring Contabo nodes for eight months. Their status page shows aggregate load but never per-tenant breakdown. Alert fatigue is real when you cannot trust the metric. I moved a client to Vultr specifically because they expose per-core allocation in their API. The graphs do not lie, but the presentation can deceive. I recommend independent monitoring with NodeExporter or similar. The graphs do not lie.