I am working on setting up Keploy at work to generate unit tests for one of the runtime component at Skit.ai
For general use cases on linux, Keploy works as below
keploy record -c "python main.py"
I discovered that on MacOS, Keploy works by containerizing the application. So it needs to work something as below
keploy record -c "docker run --name test -p 8080:8080 image:tag" --container-name test --buildDelay 30
I faced some SSL handshake issues while running it this way that I didn’t face while running the application server normally i.e. without Keploy.
After this I planned to run a linux container and run everything inside that i.e. entire dev setup, application server, keploy etc.
I noticed that I was getting the below error
🐰 Keploy: 2025-03-18T13:02:28Z ERROR failed to load eBPF objects {"error": "field K_connect4: program k_connect4: load BTF: BTF not supported (requires >= v4.18)"}
🐰 Keploy: 2025-03-18T13:02:28Z ERROR failed to load hooks {"error": "field K_connect4: program k_connect4: load BTF: BTF not supported (requires >= v4.18)"}
🐰 Keploy: 2025-03-18T13:02:28Z ERROR failed to start the hooks and proxy {"error": "failed to hook into the app"}
🐰 Keploy: 2025-03-18T13:02:28Z ERROR failed to instrument the application {"error": "failed to start the hooks and proxy"}
🐰 Keploy: 2025-03-18T13:02:28Z INFO stopping Keploy {"reason": "failed to instrument the application"}
🐰 Keploy: 2025-03-18T13:02:28Z ERROR failed to record {"error": "failed to instrument the application"}
After this, I learnt about eBPF.
eBPF is a technology that can run programs in a privileged context such as the operating system kernel. It is the successor to the Berkeley Packet Filter (BPF, with the “e” originally meaning “extended”) filtering mechanism in Linux and is also used in non-networking parts of the Linux kernel as well
MacOS doesn’t have support for eBPF. However, there’s one userspace implementation of eBPF, uBPF that I’ll try next.