We'll be building two different images for this setup:
- A
node
image that can be flashed onto the devices that are participating only in the mesh network. - An
ap
image that will be flashed to exactly one device which will act as the access points for devices wanting to connect to the mesh but that aren't part of the mesh.
Taking care of steps 1 to 8 of the following prerequesites is required for both types of images/devices, step 9 is only for the ap
image/device.
Building a deployable UAVPi image is aided by the scripts and pre-defined config files in this repository. There are a few prerequesites for starting this workflow:
- Download a suitable vanilla Raspi OS image, e.g. 2022-04-04-raspios-bullseye-armhf-lite
- Flash that image to an SD card like so:
$ xzcat 2022-04-04-raspios-bullseye-armhf-lite.img.xz | sudo dd of=/dev/sdX bs=4M conv=fsync status=progress
- Then clone this repository
$ git clone git@github.com:teschmitt/UAVPi.git
$ cd UAVPi
- Edit
pi_config/userconf_sample
in order to set up a custom default user. Generate the password hash with
$ openssl passwd -6
... enter password ...
then save the output to pi_config/userconf
:
<username>:<PASSWORDHASH>
- The self-hosted setup steps require a wifi connection, so edit
wpa_supplicant.conf_sample
and remove the suffix (remember to wipe all sensitive info from this before creating and distributing the image). - If you want to connect to the Pi via SSH, remove the suffix from
ssh_sample
. These will be copied to theboot
partition and will be read out automagically on the Pi's first startup. ./setup_medium.sh
expects the DTN7 binaries (from the Rust implementation) to be in a sub-directory calleddtn7-rs-release
. So please compile them withcross build --release --target arm-unknown-linux-gnueabi
and place them there.- Go and tell the
cross
maintainers what absolute MVPs they are. I'll wait.
Important: the following step only has to be performed when creating an ap
image:
- Edit
networking/hostapd.conf
to reflect the settings you want for the non-mesh connecting clients. Also take a good hard look atsetup_mesh.sh
starting at around line 70, where settings are getting written todnsmasq.conf
anddchpcd.conf
.
The above steps will normally only have to be done once, then you have a working environment for the following workflow.
If everything went well, you should have two partitions mounted and it should look something like this:
$ lsblk
...
sdX 8:16 1 59,5G 0 disk
├─sdX1 8:17 1 256M 0 part /run/media/<username>/boot
└─sdX2 8:18 1 1,7G 0 part /run/media/<username>/rootfs
You can then launch the script to copy all files needed for the self-hosted setup with the details from above:
$ ./setup_medium.sh --bootfs /run/media/<username>/boot --rootfs /run/media/<username>/rootfs
Since we're copying the DTN7 stuff into system directories, you will need to plug in your root password along the way. Don't forget to umount
the partitions. You can now boot up the Pi (and connect to it via ssh, if you provided the adequate credentials above).
For the ap
image, see below.
In order to install all needed software, run
$ ./setup_host.sh
This requires an internet connection and may take a while. When it's done you should see
Setup finished, you can now run the setup_mesh.sh script.
So do that:
$ ./setup_mesh.sh
This will actually muck around in the network configurations and set up all interfaces needed by batman-adv
. If you want to change any of the used options, check the networking
directory for the appropriate files.
This is largely analog to the node
image, but we need to set a flag in the setup scripts to tell them to add additional software, configs, and so on. So here we go:
Install all software:
$ ./setup_host.sh --ap-mode
This requires an internet connection and may take a while. When it's done you should see
Setup finished, you can now run the setup_mesh.sh script.
So do that:
$ ./setup_mesh.sh --ap-mode
Now for the most important part, creating the image:
- Power down the Pi and remove the SD card. Mount it on the workstation
- Find out the device ID like above
- Wipe any sensitive information (e.g. login info to your Wifi in
<path_to_rootfs>/etc/wpa_supplicant/wpa_supplicant.conf
) you don't want to distribute later on from the SD card data. - Fire up the
create_image.sh
script (this will also require yoursudo
credentials, so have those handy):
$ ./create_image.sh /dev/sdX imagename.img
It is advisable to do this once for an ap
image and once for a node
image. This will take a while, but you'll have a ready to deploy image when it's done.
The setup_mesh.sh
script will generate a hostname based on the MAC address of the wlan0
interface and write this hostname to /etc/hosts
and /etc/hostname
. The image you create will have this hostname hard-coded into it. If you want something more generic you will have to alter the appropriate files before ripping the image.
To auto-generate and set a hostname with the schema uav-<last 6 MAC addr digits w/o colons>
, simply run the autogen_hostname.sh
script.
Let's say we have two images from the above setup workflow:
- meshpi-node.img
- meshpi-ap.img
We can now flash these images to SD cards as we need them and they'll work. On first boot, make sure to run the autogen_hostname.sh
script to give the devices unique hostnames.