Newbie questions

I’m looking to set up my first node. A few things not covered in the docs I’ve been able to find:

  • I have a dedicated Core i3 box I’m planning to use. I’m most experienced with Ubuntu but I’ve used Debian before. Can I use the most recent Ubuntu Server? The prerequisites page looks like it might be out of date (or maybe I truly shouldn’t use more recent versions…) Thoughts? I can install whatever Linux is best, my preference to use something familiar is pretty mild.
  • Is it reasonable to run any other server services on the same box (e.g. a local file server) for my own use? Or is it pretty important to dedicate the entire machine to Storj?
  • Will the amount of drive space significantly impact my earning potential? I currently have a 1TB drive I could use. Would I be better off ordering, say, a 4 or 6 TB drive, and waiting to set up until I do?
  • Is there a way to test things before going fully online? For instance, I don’t yet have a UPS, but I could get one. Would I be better off waiting until I get a UPS to activate my node?

Trying to understand everything ahead of time so I can have a smooth experience. The docs seem kind of thin. Any tips appreciated!

1TB is fine to start. Remember to leave space for logs, etc. The recommendation is 90% of the drive size could be allocated to your node. (Edited, thanks Storgeez!) It will take a while to fill that. I would personally not order a larger drive, just use what you have. If that drive finally fills up and you want to commit more storage to the node, then you could consider if it is worth it. Or maybe by then a larger drive will have made itself available to you from a different retired computer. I’ve done drive-to-drive migration in Linux (Raspberry Pi OS as I’m running a RPi3) and the documentation of how to do so is straightforward.

  • Ubuntu should work fine.
  • You can use the machine for something else. If the file server uses a same drive, there might be an IOps bottleneck under heavy load which would result in loss of some potential downloads, but this would have a very minor impact on your income unless it is a very heavy load for a very long time.
  • People have reported average earnings of about $4/TB of data (combined storage + egress).
  • Online time isn’t nearly as important as data integrity. You would just need to make sure power failure does not cause any data corruption. And you would of course need to have the node turn back on after a power failure.

And, yeah, you can always upgrade to a larger drive, like Craig said. Start small and see how it goes.

Was 10% reserved space IIRC!


I agree with the previous two posts. Start off with what you have. See how it goes. Data ingress to your node will be low to start, and will increase once you surpass the vetting period. I’d say with current ingress rates, with ~900GB of space offered for storage space will take 3-6 months to fill up. At that point see where you’re at and if you’re liking what you’re seeing then maybe look into a bigger drive and migrate the data to it.

I started small with a raspberry pi 4 and a 1TB external HDD, I then migrated to a 4TB HDD, then a 10TB HDD, and eventually a 12TB HDD on that first node. After that final migration to the 12TB node, I decided to start a second node on that 10TB HDD, and so on and so forth. At this point I have 6 nodes.

Like many folks that have been running nodes for many months, I’ve cashed out some of my Storj earnings to help pay for some of the upgrades.

It is fine to use what you have. No expensive investment required.

Yes. At times there will be much HDD activity, at times it will be low. Traffic is not predictable.

Yes of course. Currently 1 TB will earn you about $3-$4 per month. So generally: Larger drive more income, but take into account that it can take a lot of time to fill up a larger drive. It is perfectly possible to start with a small drive and scale up as traffic flows in. This would be my recommandation.

Nah. I’d say UPS generally not required unless in your area power outages are frequent and lasting (but then internet maybe not be available anyway).

Ah, very helpful info, thank you. I’d imagine the most important thing on that point is to use a journaling file system?

Yeah, and depending on how often these power failures occur, you might want to do only synchronous writes. It really depends on a lot of stuff.
Well, basically, the only problem is with databases, data folder can lose some files and you won’t feel it as long as it’s a small number; databases OTOH are much less robust.

1 Like

Although that’s true, a node can be recovered even if it loses all of its databases. If this happens and if you react soon enough to fix the issue, you may miss some days of revenues but that’s about it.
Whereas it will get disqualified for good (that is banned from the network) if it loses a few percents of the data it is storing so that’s definitely the most important thing: integrity of stored files.

@o1eal: On another note, you may want to have a look at this estimator made by one experienced SNO, to have a rough idea on what ROI to expect (copy it in your own Google account and then fill in your numbers):

Really? Since when is that true? I know you can lose some of the databases but I thought losing the main one results in catastrophic failure.

So, I’m reconsidering, I will probably use a Raspberry Pi 3 with an external USB 2.0 drive. Seems like USB 2.0 could be a bottleneck but not an awful one (I’d probably want to limit the Storj bandwidth to about that speed anyway). I suppose this means I should use Rasbian, which I haven’t used before especially… but it looks like there’s a great guide here:

It almost seems like one in every 3.14 people on here run it on Raspberry Pi. USB bandwidth will not be an issue, traffic is far from it. More than likely the interface adapter in the external drive will be a bottleneck, and the drive too. USB reliability might be an issue but people run it like that.
Others will know more about the berrian you’re talking about.


I’m not sure precisely but that’s pretty old:

If you have a spare one laying around, sure. If you were considering buying something, then I’d recommend going for an RPi 4B, with at least 2GB of RAM (4GB is better).

:grin: :+1:


Did not realize you didn’t need any of the databases to pass audits. I thought piece info was stored in a database, the loss of which would result in inaccessible data. I know it shouldn’t need a database, but thought it did for some reason. Looks like I made a totally wrong assumption, thanks for the link!

P.S. I did have issues with databases before and when the node couldn’t open or access the database, it would fail catastrophically. I guess it’s technically true but if it has access to data and not to databases, the node will completely fail to start/work. Presumably deleting the databases in that case would’ve fixed it. :woman_shrugging:

1 Like

Probably :slight_smile:

You’re right in the sense that corrupted/malformed databases are an issue to be addressed and the Node doesn’t “fix itself”. One has to go through manual steps to try and fix malformed databases, and/or restore databases that cannot be recovered.
But ultimately, the good news is that a node isn’t doomed if one, some or even all of its databases are lost as Nodes are able to return data queried by clients and satellites just by looking it up on the disk arborescence organized in a clever way.

And that I think, is an amazing design feature from StorjLabs, as it makes Nodes less prone to disqualification (if the issue is detected in time) :+1:


So, you have no redundancy… if your 12TB disk breaks, you loose a really fat node…
Is this wise?

correct, none of my 6 nodes have redundancy. It’s been discussed on this forum many, many, many times, redundancy vs no redundancy. Ultimately, it comes down to user preference, but the official recommendation is 1 HDD per node. Also, just for your reference, that first node of mine has been running for almost 20 months, and has earned somewhere around 1,800-2,000 storj tokens. Some I’ve sold to pay for some new hardware, but most I’ve held, so with the current token prices, it’s earned way more than I ever thought it would, and if it died tomorrow I wouldn’t be mad.


I see. Well, I’m using a raid 1 because that is what I had available in a NAS. For sure, It makes no sense to do it on purpose for storj, but I never thought about setting up a PI…

Redundancy makes sense for storage you keep other data on. For Storj you’re better off running on each drive separately as you’ll have more storage and earning potential. At least in theory.

Yeah, I think having two single disks that can be filled is better than one disk that is mirrored. (Or whatever combination of RAID you want to run) If you lose a disk, you’re still making money on the other.

1 Like

I am really pleased to read all the answers – helps me a great deal in getting my bearings.

I’ve come to a decision on hardware, I think I have a better option than the Pi, and I have another use for the Pi anyway. I have a 2007 Mac Mini (model 2,1) with Lubuntu LTS installed, but going unused. Since it lacks HDMI, USB 3, or even an SSD, it’s not good for much else; but, I have also freed up a 4 TB external drive I can connect to it.

I’ll probably get to work in earnest setting this up as a node in the next few days.