Following this thread. Currently on RackNerd $10/year plan in LA, no timer nonsense, just "out of stock" when actually out. Performance is what you expect for $10.
This host's 'anniversary sale' is the same price as last month
LichunLars said:
Their London stock is always "waitlist"
Confirmed. I tried to order Contabo London 2GB last Tuesday, "expected availability 2-3 weeks." Nuremberg was instant. They're oversubscribed in premium locations and using the "sale" to funnel people to German DCs with excess capacity.
Manchester to Nuremberg is 22ms. Not terrible but why pretend London exists?
kate3
Member
james37 said:
RackNerd $10/year
RackNerd's $10 plan is KVM though. 768MB RAM, single core, but actual virtio-net and your own kernel. I run WireGuard (https://www.wireguard.com) on mine in Amsterdam, no issues.
The honest "out of stock" is refreshing. No psychology, just inventory.
virsh list --all | wc -l: 47
I am buy RackNerd now, thank you kate3 and james37
The website say "out of stock" very honest, I am refresh, still out of stock, I am wait
Better than red timer make heart attack xixixi
My english improve today
kate3 said:
RAID-60 SATA with a hot cache layer
Do you have a source on this? Not doubting, want to read more. Contabo's specs page just says "NVMe" with asterisk pointing to "in selected datacenters."
kate3
Member
liteltd said:
Source on this
No public source, own testing. I ran fio with --runtime=300 on their 4GB plan in Nuremberg. First 60 seconds: ~50k IOPS. Minute 2-5: drops to ~8k. Pattern matches cache exhaustion, not thermal throttle.
I asked support directly: "What is the underlying storage configuration?" Response: "High performance NVMe SSDs in RAID." No denial of tiering, no specifics.
Draw your own conclusions. I draw "don't run databases here."
virsh list --all | wc -l: 47
From New York: Contabo's NY datacenter is actually in St. Louis per traceroute. 35ms to me, not the 5ms you'd expect. They list "New York" as a region, not a city. I used https://ipinfo.io to check.
Vultr's New Jersey is actually New Jersey. 8ms. Worth the extra $3/mo to not lie to myself about geography.
srelet
Member OP
pure5 said:
Actually in St. Louis
This I did not know. My three months uptime data is from "New York" region. Let me check...
Traceroute: last hop before my VM is 209.126.2.x, ARIN says St. Louis. Contabo's own IP allocation.
Updated transparency:
- location: misrepresented
- timer: fake
- storage: tiered undisclosed
- outage: unexplained
I'm migrating off. Not because $18 is wrong, because cumulative dishonesty is exhausting.
I compile chromium for relaxation. 128GB or nothing. These $18 discussions are cute but irrelevant to my life.
That said, Hetzner's dedicated auction had a DX152 last week, 128GB DDR4, €64/month. No timer. Just inventory and honest depreciation curve.
Watts are the real currency. Contabo's OpenVZ nodes are probably E5 v2 era, 120W idle. My Ampere box is 80W under full load.
Keysales guy here. The real thing nobody mentions: Contabo's TOS has something about banning "crypto mining" broadly defined, from memory. I run BOINC for science, got suspended in 2022. Appeal took 11 days. "High CPU usage" was the reason given, not even mining.
Hetzner also bans mining but their "high CPU" threshold is documented: sustained 100% for 48h on shared vCPU. Clear rule, fair warning.
Contabo's ambiguity is the product.