ecc docker usage

docker usage

The only file you will need to write is:

your_program.bpf.c
your_program.h  # optional, if you want to use ring buffer to export events

after that, simply run this:

$ docker run -it -v /path/to/repo/:/src ghcr.io/eunomia-bpf/ecc-`uname -m`:latest # use absolute path

you will get a package.json in your root dir. Just run:

$ sudo ./ecli run package.json

to start it you can download ecli tool from eunomia-bpf/releases, we have pre-build binaries for linux x86. Small and No dependencies, besides glibc and glibcxx. Or just run this:

$ wget https://aka.pw/bpf-ecli -O ecli && chmod +x ecli

The eBPF compiled code can run on different kernel versions(CO-RE). see: github.com/eunomia-bpf/eunomia-bpf for details.

container image

simply run:

$ docker run -it -v /path/to/repo:/src ghcr.io/eunomia-bpf/ecc-`uname -m`

Or you can do that without a container, which is listed below:

Github actions

Use this as a github action, to compile online: see eunomia-bpf/ebpm-template). Only three steps

use this repo as a github template: see creating-a-repository-from-a-template
modify the bootstrap.bpf.c, commit it and wait for the workflow to stop
Run the ecli with remote url:

$ sudo ./ecli run https://eunomia-bpf.github.io/ebpm-template/package.json

Notifications

We use the same c ebpf code as libbpf, so most libbpf ebpf c code can run without any modification.
Supported ebpf program types: kprobe, tracepoint, fentry, we will add more types in the future.
If you want to use ring buffer to export events, you need to add your_program.h to your repo, and define the export data type in it, the export data type should be a C struct, for example:
```
struct process_event {
    int pid;
    int ppid;
    unsigned exit_code;
    unsigned long long duration_ns;
    char comm[TASK_COMM_LEN];
    char filename[MAX_FILENAME_LEN];
    int exit_event;
};
```
The name and field types are not limited, but we will prefer use standard C types. If multiple struct exists in the header, we will use the first one. The feature is only enabled if we found a BPF_MAP_TYPE_RINGBUF map exists in the ebpf program.

Share on Share on