Threefold Ops team looking for NVMe nodes

#1

Dear farmers.

Since we continuously have to expand the backend that is required to run the Grid and all related services, we are in search of ZOS nodes with the following:

  • 2TB+ nvme nodes (preferably 2 x 2TB or more)
  • descent modern Intel/AMD CPU
  • at least 64GB of RAM
  • at least one public IP + descent reliable internet connection

We are looking for ZOS nodes to deploy full blockchain nodes for BSC, ETH (test and mainnet), Polygon and BTC (main and test). BTC nodes are not that intensive but the others are. These nodes generate quite a lot of traffic so the farmer will get some nice extra rewards out of this on top of the public IPs of course. These workloads are also quite IOPS hungry, hence we require NVMe disks. The difference between sata SSD and NVMe is huge when you need a lot of IOPS for a chain node.
Depending on the specs, we can also rent your machines as dedicated nodes.

These nodes will serve the current bridges, development work and all the juicy new tech coming soon!

If your interested drop your farm ID here or send it via PM. Tell us a bit about your farm, location and available hardware.
If you have questions, shoot!

Regards,
TF Ops team.

2 Likes
#2

Exciting! Can you elaborate more on decently modern CPU’s?

#3

I am interested. I have the following running:

  • AMD Ryzen 9 3950X
  • Corsair Vengeance LPX 128 GB Ram
  • PNY XLR8 CS3040 2TB M.2 SSD
  • Stable internet with high speed
  • Location is The Netherlands
  • Doesn’t have public IP, but can be added.
  • Farm ID: 2278

Let me know if I can help and what I should do to help.

#4

Specific CPU models are most of the time not specified for blockchain nodes. Core count seems to be mentioned more. A minimum of 16 cores, preferably 24 and optimal would be 32. It also differs a bit between chains.
Mobile CPU’s are generally not recommended.

CPU has never been a bottleneck for us in the past, since we always allocate enough cores. Same for Memory and bandwidth. It always comes down to IOPS where the answer then becomes: NVMe :wink:

So any mid to high spec’d modern CPU will do, modern because of the encryption capabilities of modern ones (nice speed boost for specific tasks).

#5

Thanks for your reply @jefke1

Specs look great! Only that we definitely need around 3 to 4TB of NVMe storage and at least one public IP. If you could fix that we would love to use the node!

Is your internet connection symmetric btw?

#6

Dam, i have everything you need beside the public ip :frowning:

#7

There are some workarounds to run these chain nodes without public IPs but we rather not go there since they introduce a lot more management overhead.
What’s your farm ID? We can have a look at your nodes and consider using them.

#8

Nice to hear, I have 2 servers with these specs but by now no public IP, but I can activate publics anytime.

Internet stable 300/30 mbit/s down/up

Farm: 403
Nodes:

image
image1413Ă—487 30.3 KB

image
image1404Ă—489 29.7 KB

1 Like
#9

Damn, big nodes there.
How quickly could you get public IPs going? Once they are there we can deploy quickly.
I also noticed only node 5910 is online. Are you managing the other nodes with the farmerbot?

#10

Hello,
I’m a big user of the grid. I have 100+ nodes running on a daily basis.
I was a farmer for about 1 year, but unfortunately i had to quit as I needed the hardware for another projects.
My farm is/was located in Portugal and I have public IPs available. I have10G symmetric internet connection and I could setup a v1/v2 DDR3 or a v3/v4 DDR4 server if there is any interest/need. I also have about 4TB (maybe 6 TB) of Nvme that I could also use for this…
Please let me know if my location/setup could be useful for the project…

#11

But they have not 2x4TB or 4x2TB SSDs its with one 8TB SDD each.
The CPUs arent the most modern ones i think both are with 4 x Intel Xeon E7-4850 am not sure right now.

Publics are only a call far away and a bit networking…
But if i do this i have to check other things too.

How is it about SLA? by now i don’t have an UPS or failsave networking if anything happens.

Best is your services run on many servers, if one fails.

//edit: and yes they are managed with farmerbot right now

#12

Hi, so most of my nodes are nvme only the latest node in my farm uses SSD. The farm ID is 328. I am using the farmer bot. Most of the nodes use an E5-2690V3 or E5-2699V3 or E5-2697 v3. I do not know if that is powerful enough. The only issue is there is no public ip.

#13

Glad to hear that! Hope you have a good experience on the grid.

Some v3/v4 DDR4 nodes would be nice yes, especially with the 10G symmetric line. And with public IPs yes indeed. 4TB NVMe would be nice, 6 even better. Let us know once you have a node ready so we can test it.

#14

NVMe size is not important, as long as the total is around 4TB (or more). We add virtual disks to the vm (that reside on separate NVMe SSDs) and btrfs raid them inside the vm.

Xeon E7-4850 is not that new but still has lots of cores and good passmark score. The only way we can know for sure is try it. They might be too slow for BSC, but maybe good for BTC.

Don’t order too much public IPs, depending on the amount of nodes and internet connectivity we can decides how much vm’s we deploy at your location. 2 or 3 would already be good.

Unless your a gold farmer, no SLA is in play here. We run multiple of these chain nodes on the grid so we can afford to loose a few and still be operational. Everything we run is fully monitored (Prometheus/Grafana) so we know the moment a node goes down somewhere.

1 Like
#15

You seem to be in the US? Which would be good since we do not have a lot running there currently.

CPU’s should be fine, but we should test to be sure. Plenty of ram and NVMe space from what I can see in the dashboard, nice.

Would it be possible for you to setup some public IP?

#16

Can you please share what is the rate payed to farmers in IPv4 traffic in “normal” farming and in “gold certified farming” per TB of traffic?
It’s really a coincidence but today I am moving most of my gear to a DC. Not sure if I’ll keep my current 10G connection at my homelab… I am aware that BTC node don’t generate much traffic, not sure about the other blockchains…
Do you have an estimate about total traffic for this nodes?

#17

Last I knew it was about .005$ per hour it’s rented and .03$ per tb.

Best way to maximize rewards for this would be to have a node they use a dedicated so you get the bonus earning for reserved. My ips never came anywhere close to offsetting their monthly cost

1 Like
#18

and furthermore, how is the extra reward for the load on the servers calculated? on the https://dashboard.grid.tf/calculator/simulator/ it seems it calculates only idle static ressources.

#19

That is not compensated for under v3. It will be under v4.

2 Likes
#20

Unfortunately I am not able to get public ip at the moment. My isp trying to charge me a little to much for some public ip.