New node operator, have a few questions

First of all let me say that I am excited to have discovered this project. I am a decentralization freak and I work as a sysadmin/devops dude during the day. I haven’t yet, but I plan on actually using Storj for some personal projects of mine to try it out.

Anyway, right now I have two physical nodes running (redundant power, redundant networking, dual CPU, 20TB total shared of RAID 5 arrays) that are also mining XMR with half the threads. I also have symmetrical gigabit fiber with virtually no downtime. These are hand-me-down machines I have collected from work that I was already mining with. I was just curious if there was anyone else out there running “high performance” configurations such as this, and what sort of usage results you have seen. Also, does the network actually measure my bandwidth or does it just look at aggregate data?

I saw elsewhere someone mentioned it being an issue that both nodes are sharing one WAN IP. Is this a hard case or can my bandwidth/uptime overcome this? I suppose you wouldn’t want to host an NFS mount.

Just curious is all. I didn’t like seeing those bays just sit empty… I’ll run the nodes until the boards are dead

Hi and welcome.

Let me start with this: For running a node you don’t need: Redundant power, redundant networking, dual CPU, 20 TB of free space or RAID.

Nodes can be run on a Raspberry Pi or similar devices (I usually mention this device here as it is close to being perfect for running a Storj node: Odroid HC-2)

Much more important than the computer specs would be the ip addresses. The Storj satellites consider nodes on the same /32 /24 net as one node and share the traffic among them. So if you can get a different second ip for your second node it could be beneficial.

Just out of curiosity, what CPU are you mining XMR with? And don’t get me wrong: There is nothing wrong with a powerful computer that can be used for additional work besides running a Storj node. In fact, it is a good idea for utilization. But only for Storj, relatively lowend hardware is good to go.

Just to give you some idea: We saw calculation that it will take around a year to fill a 8 TB drive with Storj data and currently it seems that around 16-18 TB is max space occupation for one IP. Storj is not mining, it is storing of customer data in a distributed fashion. Low customer usage results in low network utilization and there is nothing a SNO can do to overcome this.

3 Likes

Interesting. I would like to point out I am not looking to profit really, I just like participating in decentralized projects, and if I make some tokens along the way that maybe appreciate, great. Most of my bandwidth is unused anyway, may as well put it to work for something cool. I may actually end up a customer myself.

The hardware I get is overkill. But it was free and can potentially save the systems from going down while I am away from home for work which is a lot of the time. I would rather not have to intervene with them if I can delay it at all. My main curiosity really was how much the bandwidth was weighted, but I get the customer usage part. I am still very new, but so far it looks like this team knows what they are doing. I am rooting for Storj!

Both systems are some dual-socket Sandy Bridge Xeons. I know they aren’t super power efficient, but I simply like supporting XMR (I have some Ryzen rigs too). I also run a network node for XMR.

This is all also just an excuse to have stuff to tinker with in my homelab.

Then you should definitely check that your automatic Storj updates are configured correctly and working.

I am on a DSL 100/40 line and it don’t get saturated by Storj at all by a single node.
And yes, it is all about real customer usage.

I can relate to that… :grin:

1 Like

It’s only a matter of time before stuff like Storj sees more adoption imo. I know many people including customers of mine who have censorship concerns with the major cloud providers. Decentralized storage is a step in the right direction here.

I haven’t done all my research, but a file sync service/client could be a big feature?

on the same /24 subnet. (fyi /32 is the only one IP).

https://docs.storj.io/dcs/how-tos/sync-files-with-rclone
There is also a rclone mount command to mount the bucket to a folder on your disk.

1 Like

Yes of course, my bad.