Member
ZFS on these cheap boxes is pain. Contabo kernel is old, ZFS module missing, dkms fails.
If you're benchmarking storage, note filesystem. ext4 vs xfs vs ZFS matters for the numbers. I can run same yabs on ext4 then ZFS if people want that variable isolated.
Running now. RackNerd LA to London: 95 Mbit. To Miami: 210 Mbit.
"Premium network" my left foot. But $4/mo so honestly who am I yelling at.
GreenCloudVPS Singapore though: 680 Mbit to same London target. Same price tier. Wild variance.
This is precisely what I meant by market working. RackNerd at $4 is telling you something with that 95 Mbit; GreenCloud at comparable price with 680 is telling you something else.
The revelation is not fraud. It is segmentation. You are not the segment RackNerd built that SKU for.
I have no horse in this race, just watching. In Lima our options are thin. Contabo Singapore is what I use for latency to Asia, Contabo New York for US.
Question already covered? How do I run yabs safely. Saw fail2ban mentioned. That's it?
No. Yabs calls iperf to public servers. Those servers see your IP. Your provider sees sustained outbound to iperf endpoints.
If you're paranoid: run disk test only. Network test to your own second box. Slower, less useful, but zero external footprint.
Also: change SSH port before you start. Scanning spikes during benchmark windows. Obscurity is not security but it reduces log noise.
InterServer result is here. 1 CPU, 1 GB, Secaucus datacenter.
CPU : Intel Xeon E5-2680 v2 @ 2.80GHz
Disk: 340 MB/s read, 280 MB/s write
Net : 520 Mbit to London, 410 Mbit to Singapore
Price: $2.50/moNot bad for small server. @spacelet you say bottleneck? Script run fine, 4 minutes. Not fast but work.