I am just starting out hosting a node and I am looking for some input on how I should continue.
I started out running it on a laptop I had lying around with 1.6TB of available storage, after 2 weeks Storj has consumed 300GB on it.
I have a couple of servers lying around and created a new identity on one of them, my question is, should I leave the node running on my laptop until it has filled out the remaining 1.6TB, then move it once it gets close to filled?
Or do I just move it right away?
The server I picked out was the quietest and least power hungry, an HPE DL380 G8 with 48TB HDD storage.
My second question is whether I setup the node stating I have 48TB (Might setup RAID 5 or 6), while only having a few drives connected to reduce power consumption, or do I incrementally keep restarting my node and updating the available storage pool as it gets closer to being full?
Third question, the other servers were an HPE DL385 G10 and an HPE DL360p G8 and an old Dell PowerEdge R710, do you think they would be better suited as nodes or potentially connecting the internal SAS controllers to my HPE DL380 G8 and just leaving the other servers in POST forever, their only purpose to power the drive bays that connect to the G8.
You will not get more data, on more nodes, traffic will split between, All nodes behind /24 network will be seen for system as one node. So if you will have lot of nodes, you just will spend more electricity.
If you’re alright with your laptop being online 100%, then I’d highly suggest keeping your node on it for as long as possible. It will spend a lot less power, than your server will.
When your laptop fills up, then build a new identity on other hardware - or ideally just upgrade the drive on the laptop to a bigger model or add another harddrive to it.
You should take note of official recommendations; read the entire FAQ and guides.
The main takeaways:
1 node per drive and 1 drive per node.
all nodes behind the same /24 subnet share the ingress, so they are getting traffic as a single node.
cpu dosen’t matter, only RAM and electricity draw.
first month is a bit slow because you are in the vetting period, and you get less traffic.
you must keep the node online as much as possible; after 4h of downtime, you start to loose data; after 30 days of downtime you are permanently desqualified (not you as a SNO, your node).
use UPS, avoid SMR drives.
linux + docker + ext4 proved to be the no.1 pick.
the laptop drive will fill in max 3 months; better move it sooner. There are millions of files and can take like 1 day per TB to move.
use log custom info mode; search for the thread. In defaut info mode, the log will get huge very quickly.
You can search for threads like optimising ext4, my docker run command, what can I delete on storagenode, custom log level, etc. to learn more and improve your node maintenance skills.
If these servers are online — then move the node any time. If they are not online — don’t bring them online just for storj. Unless electricity is free it won’t make financial sense. Storj is intended to take advantage of the already online but underutilized capacity.
48 Tb is far too much. It takes decades for this to fill up. My oldest node is exactly 24 months old and is filled with just under 6 TB. It also doesn’t fill up because deletions take place over and over again. I’ve been stuck at just under 6TB for about 6 months. Start small with the laptop and you’ll see if it’s profitable for you.
I will keep using my laptop and move the data to 2 of my 4TB disks until those are full, or I see if it is not worth it. Then connect those drives to the server if after the vetting period it increases enough to warrant having the drives inserted.
