I have a raspberry pi with 15 disks, summing up to ~70 TiB. While I was plotting, the farm lay out around the room, in “open air”. Needless to say, it was annoying everyone in the house, myself included.
When all was done, I bought and assembled an IKEA Platsa shelf and put 3 shelves in it - each about 10cm in height. There I placed my farm, only to see it’s failing misserably
My card in Pi is somewhat small, so I use one of disks spare space (~66GiB) for storing .chia folder.
Every couple of hours, this drive fails and I cannot unmount it (even with -f option), requiring pi restart.
I suspect this to be correlated with smaller space and heat, but can’t really measure it. Can someone share their experiences w/ me? I saw that people are using fans, but always figured I don’t
At this point, I spread out the drives in three shelves (check pic) but, other than getting some fans, this is all I can think of.
Disks may fail due to heat, if you keep them on at >=50C.
Disks may fail due to static charge buildup, if their electric circuit is not properly grounded and they are located on a non-conductive material like plastic, glass, ikea shelf
A cost-effective solution may be getting a wire shelf in Walmart. It is metal, so you can easily ground it. It will not constrict air flow - so the heat will dissipate with convection - or you can place a box fan blowing on the entire rig
Here’s a weird part - I’m actually trying to figure out what changed. All these disks were on wood before I moved it here (wooden parquet floor).
Walmart is out of reach as I live in Europe / Switzerland, but something alike can be found easily. However, being bare metal is quite important I guess and many stores sell these things with at least some protection for metal.
As explained in the video, static electrical component damage can occur in different ways. One of them is gradual degradation of function, which cannot be observed initially.
Hmm, I find it hard to imagine that the drives will just start failing simply because of a little extra heat. That being said, a few PC fans on each shelf might help to circulate some air on those shelves.
Another issue might be the electrical connections. You might want to try re-seating the physical USB connections just in case anything might have come loose.
Lastly, when you moved the drives around were they powered and spinning? I get the feeling that these drives are very sensitive to movement and intolerant to knocks when they are powered. You might want to run a smart test on the drives to make sure the platters haven’t been damaged during the move.
I tried to make sure they are not spiining, but I could be a bit guilty. I actually only made sure PI was off, but actually - as you write about it - I figure that the USB hub they are connected to has its own power source and they did blink as live… I figured it’s not spinning so it’s fine…
another ultraweird thing is the fact that all is good while chia node is syncing, all seem to be good. only once it’s live the problem starts occuring. right now it synced and I started making a copy just to have a more recent db backup. and no issues at the moment…
I should also point out that the USB3 connector on external drives are very fragile. They have been known to be damaged by pulling on the cables in a bad way. If any of the drives are misbehaving I would power the lot down and inspected the connectors on the drive side to make sure the USB socket hasn’t broken from the PCB. You can do this by giving the USB plug a gentle wobble.
It seems that situation is a bit more stable now. For 2 hours running without a glitch.
It could be that one of the drives is creating issues. I mean, one or more drives on a hub which is currently off.
… a question on this matter - I still don’t know why would (in case it is one of the drives in the mentioned hub) one drive cause ALL drives to be unmounted (which actually happened - something like this was in the dmesg).
Im farming on a pi too with 64GB and my DBs also on an external USB-HDD without any problems so far ( was rather skeptical if it would work ).
There are three differences about our setups:
I only linked db-folders, but not whole .chia-folder.
Im using only 8 disks.
My drives are cooled.
After all ive read here in this forum, I consider pis with more than 12 drives unstable in most cases.
The fact, that ur problems occur after sync, when ur system starts to use all drives, also points into that direction imho.
Could they be having too much vibration? I don’t think they’re designed to lay sideways like that. I say that because on the bottom of your drives sold be some rubber that serve as anti slip / absorb some vibrations.
at this time, it looks like it’s pi 4 after all … quite sad if it’s so.
situation at the moment:
PI: 6 drives (until half a hour ago, it was 5 disks on 50% filled USB hub).
Laptop (new addition): 1 ub with 10 disks. 4 are on the floor, 6 in my ikea platsa :)… I’m still not giving up. This looks like a stable config.
20 hours later and there hasn’t been a single glitch. it could very well be that my PI4 cannot handle these usb hubs / so many disks.
in the meantime, i have my plotting machine running since I added a graphics card in it when ordering - it’s now in parallel digging for BTC (actually, my goal was ETH, but something went terribly wrong so i’m going w/ the flow). i wanted to have this machine potentially farming as well, but all my disks have been formatter w/ XFS.
so… how to fit 16 disks on 1 PI?
and why not (except the obvious - it’s not working for me)
I don’t know why you can’t farm many disks on Raspberry Pi 4. I am farming with 20 HDDs on my Pi without any issue. And there’s also one time I can farm with 21 HDDs.
And if you also used a USB-C port on the Pi, you could go with 30 drives+ on a single Pi (there’s someone on this forum showing this method but I can’t find the post).
You can check your hubs to see what could be the issue.
