Hi there, i was a former node operator, years ago. I want to come back to what it! My question is: is it possible to run a node with 3HDs merged (only one) on RPi2? If so how?
Hello @toma ,
Welcome to the forum!
You may try. However, it’s not recommended to use merged disks (spanned, mergefs, RAID0, JBOD, LVM/zfs simple pools, etc.): with one disk failure the whole node is gone.
The key factor would be a memory, the node can use up to 800MB (usually 80MB-150MB) with high load and normal disk subsystem on ext4, so if you have only 512MB on your RaspberryPi 2 then it could fail with OOM especially if your disk is slow (the node will start to use much more RAM in this case).
You also need to use a headless OS to do not use too much RAM, and it also should be capable to run docker.
See:
However, please note - some tuning settings from this guide cannot be applied to rpi2, they are applicable only to pi3 models.
Does anyone have a guide on how set it up? The model has 900mhz and 1gb ram! I’m searching for it without success!
I “think” the CPU on the rpi2 is too many revisions behind. I looked into this about a year ago and I recall it not working because of this. If you have an rpi3 it will work fine, but I believe a 2 is not compatible.
2 Likes
I guess it would be possible to cross compile for the target cpu, perhaps with the older version of go, but likely won’t be worth the efforts. You can buy rpi4 for $75 from Newark today, or get any of the zoo or clones. Some even have a pair of sata ports, which can help make in into an interesting appliance for storj: disk + SSD.
1 Like
You may try to setup a node using the guide for Raspberry Pi3, provided above. There is no specific guide for Raspberry Pi2 and unlikely will be, because nobody uses it or able to setup. You may be a first one who would try.
Please report back about your adventure, and in case of success we can make a such guide together.
3 Likes
I have a Node on a Pi 2 which runs fine, i simply used the guide for the Pi3
(On 3. Dec the Node was not running because of network issues)
4 Likes
That’s good to know. The CPU in the Pi 2 is the same in the Pi 3, just slower. I would think the 1gb of RAM would be an issue, but you got it working so good deal.
3 Likes
Even with the 1GB of RAM I’ve got 3 nodes running well on a single RPi3, have been doing so with Docker since early 2021. More than 1 disk might be possible on an RPi2? I do know there are some quirks I need to get around to digging into. Example, I just checked on them and they all show as only having 54 mins of uptime. I know I didn’t restart them and they didn’t get an update pushed so something tripped them into a restart. But restart they did apparently, and are looking good.
I might have an RPi2, HDD, and an enclosure kicking around. Maybe time for a 4th node as my 3rd RPi3 node is finally getting close to full!
If you are so low on resources, why are you wasting them on docker?
Sorry but you are all late, my nodes are running on RPI4! 
Because that is what was available at the time I stood them up. They’ve worked generally without much issue so I haven’t bothered to look into the native Linux install.
1 Like
You never know, this thread could be useful for some other SNOs 
Well now that you’re on RPi4 you’ll have plenty of resources and feedbacks available to you, as many run Nodes on RPi4.
I used to run one (4GB of RAM) with 4 main nodes (and 4 disks) on it.
This is likely an OOM killed them, but it’s also possible that you have FATAL errors in their logs. 1GB of RAM could be not enough for OS and three nodes, even if the nodes itself usually do not use more than 150MB of RAM each.
3 Likes
Probably, I know once the nodes are started the CPU calms down and only the memory seems to be crushed. I do need to look into it at some point but if after failure/halt they’re restarting on their own and there are times that they will run without restart for weeks at a time I’m not terribly concerned about it. Maybe I’d win more egress races but honestly for right now I’m content just to let it chug away with very little attention on my part.
When I commit time to it again I’ll probably check out the native Linux client and ditch Docker. Might resolve those restart issues altogether.