So I see that I currently have 4.06TB of data stored. In my payout statement for Feb says “Disk Average Month: 3.15TBm”.
What exactly does that mean and why does it not match my total stored data? It has always been a little behind since I started the node.
TBm is the TB average over the entire month.
If you have 0 GB stored on the first day of the month, and 1000GB stored on the last day of the month, TBm will be 500ish.
As a rule of fumb, TBm will always lack a bit behind your total stored 
Ah okay thank you. I thought something was wrong with my node or something haha
Nah no worries, you’re good mate.
As long as your online scores are good, and your node is not restarting all the time (because then the file walker won’t finish)- you should be good.
You also have to take into account that it changes depending on the day of the month, for example, right now I have 3.47TB but in reality, I have about 5 TB stored. It will increase with the days.
Yeah all good on that. Lowest I have for online score is 99.86% and my node has been running for nearly 500 hours now.
I keep seeing that in posts a lot. What is the file walker?
File walker is a software that periodically checks all of the pieces on your node, and compares them to what the satellites think you have.
It’s much more complicated and involved than that, but I don’t personally understand file walker well enough, and even if I did, I don’t have the technical know how to explain exactly what it does.
Perhaps @elek can?
TBm is very easy, it is 1 TB stored for 1 month, if you hold 1 TB only 15 day it will be 0.5 TBm
See the main posts about the suject:
https://forum.storj.io/t/tuning-the-filewalker/19203
https://forum.storj.io/t/disk-usage-discrepancy/24715
https://forum.storj.io/t/debugging-space-usage-discrepancies/24874
Actually it would be 1000GB/30 ~33.(3) GB/mo ~0.03(3)TBm if it’s stored only one day.
The same as kWh, the used storage is calculated as GBh. 1TBm = 1TB stored for the entire month.
See an explanation there:
Interesting. Why would someone want to slow down the filewalker? I’ve seen someone ask about it before.
Because on a weak systems it may affect uploads/downloads. The disk has a fixed amount of IOPS, and hardware may be a limiting factor too, so running a heavy job (scanning millions of pieces on the disk) could be a challenge for a weak hardware, so we implemented a lazy mode.
But it doesn’t work well anyway, so to speedup a process you may disable this mode and run the filewalkers as before - with the same priority as a paid load (uploads and downloads).
So weak systems like a pi 4 or maybe SMR drives?
Currently I run a pi 4 8gb with an external USB CMR drive and get around 2TB bandwidth a month on 500 down and 21 up internet. I believe this is normal?
Pi2 and Pi3 (0.5-1GB RAM accordingly)
Pi4 is much more powerful.
Any usage is normal
Gotcha, I didn’t know how low power a Storj node could be. I also have my pi running on an SSD via USB.
One question I have been wondering about too is, I hear SMR drives aren’t very good as Storj nodes due to the shingled layout and process of overwriting and rewriting. I do have 2x additional 8tb SMR drives collecting dust. Is there a way to somewhat make those into a hybrid with an SSD or something to make them work better?
It can be run even on OpenWRT router, see Running node on OpenWRT router?
unfortunately - this is - bloody true…
Likely yes, but you would be forced to use either Primocache (Windows) or advanced FS like ZFS to allow you to use either RAM or SSD.
So probably way more complicated than it’s worth to get some drives running well enough. Can they be used usefully in any way? Redundancy?
I do not get it, sorry.
You always can run one node per HDD and it will be ok. I mean - no RAID, no expensive FS like ZFS or BTRFS, just a plain ext4 and your nodes will likely be ok.
I mean like use the CMR drive as the node and the SMR drive as a file backup just in case the CMR node drive dies and loses data. Is that a good idea or even possible?