You are viewing a single comment's thread from:

RE: Announcing my Hive Full API Cluster: hive.roelandp.nl

in HiveDevs4 years ago

Great work!
It is wonderful to see decentralisation progressing more and more on Hive.
I regularly run the Peakd node speed test and look forward to seeing yours on the list (its not there yet).
As I generally use api.pharesim.me (although its in error right now) yours will be a close competition for best node for me.

I'm wondering what CPUs you are using.

I have a 128Gb i7 6800k server that is quite under utilised and parts to make another 2 x 64Gb servers. I also have direct fibre connection.
image.png

How hard is it to set up and does every machine have to be completely dedicated?

Sort:  

Hey A,
Re: not listed: I had a Jussi config error, was just testing with asgarth and fixed that now.

As hiveD runs single threaded the processor cores don't matter that much afaik. Processor speed helps with replay times and num threads + processor speed with compiling duration. More cores helps with more workers for hivemind and nginx processes.

1x 128GB might be a bit daunting for the task as this setup runs one account history node (non MIRA) using TMPFS and then another hived-node in MIRA config. I have the hetzner entry level AMD processors. If you have 2 servers like you mention I think you will succeed, based of my conversations with @gtg :) . My servers are configged with NVMe disks in Raid0 which helps a lot with writespeed.

Thanks. When I get some spare time from legal stuff re HF23 I will look closely at this.
BTW: We (@jpbliberty) are helping a number of victims of HF23 with getting their stolen property returned. If you are interested you can contact me on [email protected] (protonmail encrypted) and keys on open pgp.

[email protected]

im interested i will try to not forget to reach out. But count me in.