In this post, I’m going to show you how to build a complete ethereum node and wallet using a raspberry pi 3. Why? Because running an ethereum node allows us to: contribute to the Ethereum network; store our wallets in a private and secure location; and it can be useful also when experimenting some kind of crypto-algorithmic trading (you can generate a new address from code, being sure that’s synced and stored in a private location, do transactions and so on).
First of all, the requirements list (with Amazon’s links!):
- Raspberry Pi 3 ~ 38 € / 45 $
- Case, power supply and heatsink kit ~ 14 € / 17 $
- Very cheap UPS ~ 28 € / 34 $
- 2 x external HDD 1 TB ~ 2 x (52 € / 62 $) = 104 € / 124 $
- USB switch with external power supply ~ 16 € / 19 $
- 32 GB Micro SD card ~ 17 € / 20 $
For a total expense of about 217 € / 259 $
A brief explanation of why we need that:
- The Raspberry Pi 3 is the most supported ARM device around. It has 1 GB RAM, 4 USB 2.0 ports, and 1 ethernet port. It’s possible to run Linux on it; in fact there are a lot of distributions available.
- Keeping the Raspberry Pi powered on 24/7 produces heat: we have to dissipate it. This guarantees a long life to our device, plus in the kit we also find a nice plexiglass case that allows us to protect the device from external damaging.
- A power outage is the second main cause of hardware failure after heating (or better, of filesystem corruption and subsequent data loss). To prevent serious damage a UPS is required. Also, the Raspberry Pi power consumption is not that high, therefore a UPS can keep it up for a reasonable amount of time (hey reader, are you a Raspberry Pi hacker and you know how to use the GPIOs to monitor the wall outlet and trigger a safe shutdown procedure when the UPS is UP but the wall power is DOWN? Maybe you also know how to start it when to power comes back, and doing all this without any information from the UPS (no USB available)? Contact me!).
- The blockchain size increases in size day by day. A simple SD card is not enough to store the whole blockchain. Moreover, the SD card is a single point of failure and we should prevent that a failure makes us lose the blockchain and - most importantly - the wallet!
- The Raspberry Pi does not provide enough power to the USB ports. Thus, since almost every external HDD is not externally powered, we have to provide enough power using a USB switch.
- The Raspberry Pi comes with no storage, an SD card is required to install an operating system.
OK, let’s start!
I choose this one instead of the most common Raspbian because I love Archlinux. Also, it’s really easy to install and use. More importantly,
geth (the command line interface for running a full ethereum node) is already packaged and available in the community repository, therefore installing it is as easy as
pacman -Syu geth.
Installing Archlinux ARM is straightforward. The installation guide on how to create the SD card is pretty clear. I’m going to show 2 ways to create the SD card. The first on Linux, the second on Windows.
Prepare the SD card while running Linux
Insert the SD card into your SD card reader. The device
# Become root: su # Create a new partition table fdisk /dev/mmcblk0 # From the guide: # At the fdisk prompt, delete old partitions and create a new one: # Type o. This will clear out any partitions on the drive. # Type p to list partitions. There should be no partitions left. # Type n, then p for primary, 1 for the first partition on the drive, press ENTER to accept the default first sector, then type +100M for the last sector. # Type t, then c to set the first partition to type W95 FAT32 (LBA). # Type n, then p for primary, 2 for the second partition on the drive, and then press ENTER twice to accept the default first and last sector. # Write the partition table and exit by typing w. # Create a new work directory and move in mkdir /tmp/wd cd /tmp/wd # Create and mount on /boot the FAT filesystem mkfs.vfat /dev/mmcblk0p1 mkdir boot mount /dev/mmcblk0p1 boot # Create and mount on / the ext4 filesystem mkfs.ext4 /dev/mmcblk0p2 mkdir root mount /dev//mmcblk0p2 root # Download and extract the root filesystem wget http://os.archlinuxarm.org/os/ArchLinuxARM-rpi-2-latest.tar.gz bsdtar -xpf ArchLinuxARM-rpi-2-latest.tar.gz -C root sync # Move boot files to the first partition: mv root/boot/* boot # Umount the two partitions umount boot root
Now you can insert the SD card into the Raspberry Pi. Connect the UPS to the wall outlet, the Raspberry and the USB switch to the UPS, the HDDs to the switch and the switch to the Raspberry. Connect the ethernet cable too and power it on. Done! You can jump to the Archlinux configuration.
Prepare the SD card while running Windows
If you’re a Windows user, you can just:
- Download Win32 Disk Imager: https://sourceforge.net/projects/win32diskimager/
- Download the Archlinux ARM SD image: https://sourceforge.net/projects/archlinux-rpi2/
- Run Win32 Disk Imager. Select the right device, the
.imgfile downloaded and
Your SD card is almost ready. Almost, because we have to resize the filesystem of the root partition, otherwise we only have 2GB available instead of the 32 GB that our SD card allows. Now you can insert the SD card into the Raspberry Pi. Connect the UPS to the wall outlet, the Raspberry and the USB switch to the UPS, the HDDs to the switch and the switch to the Raspberry. Connect the ethernet cable too and power it on.
Download putty and connect to the IP that the router assigned to the Raspberry Pi.
Now we have to resize the filesystem of the root partition, while the system is running. To do that we have to delete the partition (don’t worry, data will be preserved) and create a new one that occupies all the free space.
# Become root su # password root # Use fdisk to change the partition table fdisk /dev/mmcblk0 # At the fdisk prompt, delete the partition number 2 # Type d then 2. # Create a new primary partition (id 2) # Type n then p. # At the prompt for the first sector and last sector just press enter. # This will automatically start from the old first sector and will set the last # sector to the end of the available space. # Write the partition table and exit by typing w. # Now reboot reboot
Now log in again to the Raspberry Pi.
# become root su # password root # Now we use resize2fs to let the kernel aware that the filesystem has been enlarged resize2fs /dev/mmcblk0p2 # logout exit
Done. Now you’re ready to configure your brand new Archlinux setup.
Our aim is to setup a full ethereum node, whose data (blockchain and wallet) is stored onto the external hard drives. Before doing that we have to configure a RAID 1 for our external HDDs.
Note: the Archlinux image contains out-of-date software. This is never a good thing. Updated software give us better security. However, we’re going to exploit this fact before upgrading the system.
ssh server installed is so old that still allows, as a default option, the remote login of the
root user. We’re going to exploit this in order to create and mount the RAID 1 partition on
/home. In this way, we can completely remove the home directory of the
alarm user (and of any other user with a home directory) without any inconvenient.
Login remotely as root user (under windows use Putty with user
root and password
root, under Linux just user
ssh with the same credentials)
Our 2 external HDDs are located under
/dev/sdb. First of all, we’re going to create a partition table on those devices and a partition for our data.
Since we’re going to store a lot of data on these hard drives it’s recommended to create a GPT partition table. For doing that, we need
# Install gdisk pacman -Sy gptfdisk # Use it to partition both hard drives. Default values are OK since # we want to create a single partition as big as the whole hard drives gdisk /dev/sda # Now create a single partition as big as the drive # Type n then p. # Accepts the defaults for every following question. # Type w to write and exit gdisk /dev/sdb # Do the same exact procedure of above # We created /dev/sda1 and /dev/sdb1 partitions # Create an ext4 filesystem on these partitions mkfs.ext4 /dev/sda1 mkfs.ext4 /dev/sdb1
Alright, we partitioned our HDDs. Now we can create a logical RAID 1 volume. RAID 1 is the most straightforward RAID level: straight mirroring. This means that if one of our drives fails, the block device provided by the RAID array will continue to function as normal.
We’re now going to use
mdadm to create the new block device
mdadm --create --verbose --level=1 --metadata=1.2 --raid-devices=2 /dev/md0 /dev/sda1 /dev/sdb1
The last thing to do is to format the drive:
Done! Our drive is ready to use. (For a more comprehensive guide, please refer to the official documentation: https://wiki.archlinux.org/index.php/RAID)
RAID /home and user setup
Since we’re still logged as
root, we can remove the current default user
alarm and its home directory. Then we’re going to mount our RAID device to
/home in order to have redundancy for our data. Next, we create a new user to use
userdel -r alarm mount /dev/md0 /home # Add the user `geth`. Automatically creates /home/geth useradd geth # Set a new passowrd for user geth passwd geth
Now, we need to make the mount of
/home permanent across reboots. Change the
/etc/fstab adding the line (using
/dev/md0 /home ext4 defaults,nofail 0 0
Save the changes and exit.
Now that we have: a new user
geth; set a new password; a RAID device that’s mounted on
/home; we’re ready to upgrade the whole system.
# First change the root password passwd # Now upgrade everything pacman -Syu # reboot reboot
After the upgrade, we’re unable to login via ssh as
root (that’s a good thing). Let’s login as
We’re now going to install
geth, configure it to start at boot and fix some little problem that using external HDDs can cause to our RAID-1 setup.
geth & startup services
geth, we have to disable any power saving options in our HDDs. We have already connected the external HDDs to an externally powered USB in order to give them enough power to work.
External hard drives slow down when the disk is not used and/or power them off. This can be a problem for our RAID configuration because once a drive turns off, it becomes unavailable as storage and so our RAID setup becomes useless.
We have to disable both HDDs power saving settings and we have to do it every time we turn on our Raspberry Pi. A simple systemd service file is what we need.
su # become root # install hdparm pacman -S hdparm
Now, as root, use your preferred text editor and create the file
/etc/systemd/system/hdparm.service with this content:
[Unit] Description=Disable HDDs power saving [Service] Type=oneshot ExecStart=/usr/bin/hdparm -B 255 /dev/sda ExecStart=/usr/bin/hdparm -B 255 /dev/sdb ExecStart=/usr/bin/hdparm -S 0 /dev/sda ExecStart=/usr/bin/hdparm -S 0 /dev/sdb [Install] WantedBy=multi-user.target
This service file will disable Advanced Power Management feature and the spin-down of both HDDs. Now we start the service and enable it on every boot.
systemctl start hdparm.service systemctl enable hdparm.service
OK. We’re finally ready to install
geth and make it run on every boot.
# Install geth pacman -S geth
Create the service file
/etc/systemd/system/[email protected] with the following content:
[Unit] Description=geth Ethereum daemon Requires=network.target [Service] Type=simple User=%I ExecStart=/usr/bin/geth --syncmode light --cache 64 --maxpeers 12 Restart=on-failure [Install] WantedBy=multi-user.target
Please note that we’re setting up a light node. This means that you won’t download the whole blockchain. This is the only mode that requires low hardware resources and that can work on a Raspberry Pi right now.
You can change the
maxpeers flag with the number of peers you desire. The highest your internet speed is, the highest can this number be (the default value is 25).
Let’s enable this process on boot and make it run from the
Done! Now the node is syncing with the ethereum network and you can use every feature of
geth (here’s the official doc: https://github.com/ethereum/go-ethereum/wiki/geth).
You can monitor the output of
systemctl status [email protected] and you can launch a console attached to the running node with
geth attach (from the
A small note:
The blockchain synchronization can take a very long time. Also,
geth uses a lot of RAM when syncing. Therefore is not unusual that the kernel kills the process and then systemd restart it. However, once the whole blockchain has been synchronized and the number of blocks to receive in an hour is low, a Raspberry Pi with 1 GB of RAM can handle the synchronization without crashing.
Ideally, if you already have a geth node synchronized, it’s suggested to copy the blockchain using
scp or any other remote copy tools instead of using
In this post, I described how to set up an Ethereum node on a Raspberry Pi 3 using Archlinux ARM. After the initial sync, we can use it to send/receive ETH, handle power failures thanks to the UPS and have redundant data to handle HDDs failures.
Please note that there are a lot of things to do to increase the security of your node. I just leave here few hints:
- Do not expose the node outside the LAN
- iptables with ‘deny all’ policy is a good friend
- Configure your firewall in the router
- Use strong passwords
- Do not install
sudo. If you install it, please add
Defaults rootpwand use a different password for
- Place the node in a secure location
- Evaluate if encryption of the RAID device is an option (although
gethforce us to protect our wallets with a passphrase, that’s enough for my needs).
If you found this post helpful let me know it! Just leave a comment in the form below.
And since we are here to try ETH, if you want to offer me a beer (or anything else) you can just send me ETH to: