The network has more nodes, more efficiently and less traffic. All together I think PIs lose the race, plain and simple.
Other machine node are simply faster than PIs and the network is a lot more faster.
Itās the Pi architectureā¦ even the Pi 4.
The USB ports are not operating at PCI bus speeds like they do on a PC.
I wonder if this could solve the problem https://www.seeedstudio.com/Dual-Quad-SATA-HAT-for-Raspberry-Pi-4-or-Rock-Pi-4-p-4389.html
A fancy USB-SATA adapter is still a USB-SATA adapter and would work at USB speed.
I order this allready and im in contact whit the Devs. So lets see what it is doing or not
Donāt. I have the storage board expansion on my PIsā¦same success rate, donāt bother.
The Raspberry 4 or 3 Version?
And which HDDs you are using?
@anon27637763 In my experiment I used the exact same disk in the exact same USB enclosure but connected to a home PC. So the slowdown is certainly not in the USB interaction.
My next candidate idea is that these extra 20-40ms are spent somewhere around the time Storj process accepts the incoming request, spawns the worker which does the handshake (or whatever it does) and starts processing the upload. Something is painfully slow there. Could it be the communication channel encryption? I guess RPi doesnāt have any hardware encryption support.
Itād be useful though to break down the whole upload request into stages to see which one is so much slower.
i think we need a special Raspberry Sateltie It should called āEco Sateliteā or something else, because the Nodes consume less electricity ;D
Think Green ^^ @StorJ
Can someone Tag a Dev, that is maybe interested in this Topic?
Maybe they are interested too about that what we try and think here.
Because its sounds to me, that StorJ is happy to have Raspberry Nodes.
And if we can optimize something or find a good was to run a Node on them, its a win/win situation for all of us. Maybe in the future there is a special Raspberry Node Script whit some chances to the normal.
What do you think, if the hardware encryption is the problem, can āgoogle Adiantumā be a help?
node not need to support encryption on node, as it not encrypt ir decrypt anything.
Okay, but when the USB is not the problem and encryption is not the problem too, can someone test the idea from xopok?
My knowleg of these is to bad, to test things like this.
But im happy tu support you if i can and maybe we will find the āproblemā.
Pi 3B+ with a WD 7200rpm. Canāt precise which HDD model it is.
But I can say that, prior to version 0.30, my PIs were booming, with a success rate over 96%ā¦both of them.
So, Iād say something changed on the Storj release as well.
Okay, i will test the new Verison of the HAT whit a Raspberry Pi 4. Maybe you have no diffrent whit the hat because USB 2.0
Raspberry 4 is USB 3.0 and ryzen bus on it work faster.
- Broadcom BCM2711, Quad core Cortex-A72 (ARM v8) 64-bit SoC @ 1.5GHz
- 1GB, 2GB or 4GB LPDDR4-3200 SDRAM (depending on model)
- 2.4 GHz and 5.0 GHz IEEE 802.11ac wireless, Bluetooth 5.0, BLE
- Gigabit Ethernet
- 2 USB 3.0 ports; 2 USB 2.0 ports.
- Raspberry Pi standard 40 pin GPIO header (fully backwards compatible with previous boards)
- 2 Ć micro-HDMI ports (up to 4kp60 supported)
- 2-lane MIPI DSI display port
- 2-lane MIPI CSI camera port
- 4-pole stereo audio and composite video port
- H.265 (4kp60 decode), H264 (1080p60 decode, 1080p30 encode)
- OpenGL ES 3.0 graphics
- Micro-SD card slot for loading operating system and data storage
- 5V DC via USB-C connector (minimum 3A*)
- 5V DC via GPIO header (minimum 3A*)
- Power over Ethernet (PoE) enabled (requires separate PoE HAT)
- Operating temperature: 0 ā 50 degrees C ambient
The USB port of the RPi is not functioning at the same bus speed as a USB port on a PC. Thatās partially the problem with the RPi setups.
The USB drive is not necessarily the problem. Itās the host controller on the RPi thatās the problem.
But it was working pretty good prior 0.30 release.
Would the controller be the only issue?
Using ānoatimeā on fstab also didnāt change anything.
This has nothing to do with the 0.30 release. The timing was coincidental with the end of a very intensive upload/download test performed by Storj. Without the glut of traffic, slower nodes loose the race. The SoC nodes are quantifiably slower than a full featured workstation/server/PC. Itās just the reality of the situation. Right now we are in an odd place where 3rd party devs are still experimenting, so not a good indication of long term traffic (fingers crossed), and there are currently relatively minimal test uploads/downloads being run by storage.
TL;DR No one should expect consistent traffic ever, regardless of hardware.
Yes, I think so to. Letās just wait and see, fingers crossed for Production and use cases on Tardigrade actually go on