camel-function

Out-of-process function runtime for rust-camel.

Provides the function: DSL step — executes user-supplied TypeScript/JavaScript functions inside Deno containers with automatic lifecycle management, hot-reload, and structured error mapping.

See docs/function-step.md for the full DSL reference.

Quick Start

use camel_function::{ContainerProvider, FunctionConfig, FunctionRuntimeService, PullPolicy};
use camel_core::context::CamelContext;
use std::time::Duration;

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    let provider = ContainerProvider::builder()
        .image("kennycallado/deno-runner:latest")
        .pull_policy(PullPolicy::IfMissing) // Never | Always | IfMissing (default)
        .boot_timeout(Duration::from_secs(30)) // max wait for container /health (default: 30s)
        .build()?;

    let service = FunctionRuntimeService::with_container_provider(
        FunctionConfig::default(),
        provider,
    );

    let mut ctx = CamelContext::builder()
        .with_lifecycle(service)
        .build()
        .await?;

    ctx.start().await?;
    Ok(())
}

ContainerProvider options

Builder method	Type	Default	Description
`.image(s)`	`&str`	—	Docker image to use as function runner (required)
`.pull_policy(p)`	`PullPolicy`	`IfMissing`	`Never`: no pull; `Always`: pull on every start; `IfMissing`: pull only if image not found locally
`.boot_timeout(d)`	`Duration`	30s	Max time to wait for the container `/health` endpoint to respond after spawn
`.instance_id(s)`	`&str`	random UUID	Scopes container labels for test isolation

Runner

cd crates/services/camel-function
docker build -t kennycallado/deno-runner:latest runner/

Endpoints on port 8080: GET /health, POST /register, POST /invoke, POST /shutdown.

Security: --allow-net=0.0.0.0 --allow-env=PORT only.

Container Lifecycle

Containers are spawned on first function registration per runtime.
ContainerProvider::Drop attempts graceful shutdown when the provider is dropped (best-effort, no panic outside tokio runtime).
ctx.stop() triggers service shutdown which stops all runners.
After stop, no containers with label camel.function.runner=true should remain.
If a container leaks (e.g. process killed without graceful shutdown), re-running will create new containers; orphaned ones must be cleaned manually with docker rm $(docker ps -q --filter label=camel.function.runner=true).

Testing

cargo test -p camel-function
cargo test -p camel-function --features docker-tests

License

Same as rust-camel.