This repository provides precompiled Fabric binaries (based on Fabric v2.1) for AArch64/ARM64 that were created as part of a project at Linköping University. They may contain issues and will not be maintained.

The binaries are distributed with a copy of https://github.com/hyperledger/fabric-samples that has been modified to use the ARM docker images.

There are also a set of patch files in this repository that contains the changes made to the source code of fabric, fabric-baseimage, and fabric-ca to make these compile on ARM They likely have to be applied to the exact commit they were made for to avoid conflicts though.

These are:

• fabric: f7f505bd76ae72526c17a255f970a0114dea19b6 (https://github.com/hyperledger/fabric/tree/f7f505bd76ae72526c17a255f970a0114dea19b6)
• fabric-baseimage: 12718aaae2e45f4c69043073292c9ff8850bbc57 (https://github.com/hyperledger/fabric-baseimage/tree/12718aaae2e45f4c69043073292c9ff8850bbc57)
• fabric-ca: 34c5a3c5223e461e8a239e30bd4420026da2fd0a (https://github.com/hyperledger/fabric-ca/tree/34c5a3c5223e461e8a239e30bd4420026da2fd0a)

To apply these changes and also update to a later version, you could probably (for each of these three) clone the repository, checkout the commit specified above, apply the corresponding patch with git apply <patch file>, commit or stash the changes, and then update with git pull. This might cause conflicts, so if you are happy with an older version you could also stop after applying the patches and compile from there.

Using the binaries and modified fabric-samples provided here, getting Fabric to run on a Raspberry Pi or other AArch64-powered systems can be done using the following steps:

1. Ensure that the target processor has support for AArch64. Recent models of the Raspberry Pi 2 (revision 1.2 and newer using the BCM2837 CPU), all revisions of the Raspberry Pi 3 and Raspberry Pi 4 should be compatible. This project was tested on a Raspberry Pi 4 4GB rev. 1.1. Fabric does not compile on 32-bit architectures due to integer overflows, which excludes the older Raspberry Pi 1, Raspberry Pi 2 rev. 1.1, and all current Raspberry Pi Zero models, since they lack 64-bit support. While it is possible that the Fabric code could be modified to work on 32-bit systems, this was out of scope for this project.
2. Install a 64-bit operating system. While several Raspberry Pi models have 64-bit support, many operating systems (including Raspbian) still uses 32-bit kernels. For this project, Ubuntu was used as it has a large set of packages available for AArch64.
3. Install Docker as specified in the Docker documentation. If you are familiar with Docker and don't want to use the fabric-samples provided by Hyperledger, stop here and use the Docker images available here.
4. Install golang and docker-compose with sudo apt install golang docker-compose.
5. Add the "ubuntu" user to the Docker group with sudo usermod -aG docker ubuntu followed by logging out and back in again
6. Download the binaries available in the release section of this repository and unpack the compressed archive. The steps outlined in Hyperledger's instructions for setting up a test network should now work as detailed below.
7. As a workaround to an issue (#1) where a non-arm64 compatible Docker image is being referenced by Fabric, download the fabric-baseos image manually with docker pull busan15/fabric-baseos:2.1 and tag it with docker tag busan15/fabric-baseos:2.1 hyperledger/fabric-baseos:2.1 to force the use of the arm64 image.
8. Change to the test-network directory by running cd fabric-samples/test-network.
9. Start the network with ./network.sh up.
10. Create a blockchain channel with ./network.sh createChannel.
11. Deploy the sample chaincode with ./network.sh deployCC. This command sometimes fails with the message Error: failed to endorse chaincode install: rpc error: code = Unavailable desc = transport is closing, although the chaincode seems to be successfully deployed after rerunning the command a total of three times.
12. If everything worked correctly, the output should include something like [{"Key":"CAR0","Record":{"make":"Toyota" [...].
13. To futher interact with the network, it might be necessary to specify the location of a configtx.yaml file. This can be achived through the command export FABRIC_CFG_PATH=$PWD/../config/, which uses the sample file provided with fabric-samples.