Trait DeploymentDigestSink 

pub trait DeploymentDigestSink:
    Send
    + Sync
    + Debug {
    // Required method
    fn record<'life0, 'life1, 'life2, 'life3, 'async_trait>(
        &'life0 self,
        deployment: &'life1 str,
        service: &'life2 str,
        digest: &'life3 str,
    ) -> Pin<Box<dyn Future<Output = ()> + Send + 'async_trait>>
       where Self: 'async_trait,
             'life0: 'async_trait,
             'life1: 'async_trait,
             'life2: 'async_trait,
             'life3: 'async_trait;
}

Sink the runtime uses to persist a service’s most-recently-resolved image digest into the deployment store, so a daemon restart can recreate the service from the already-local image (by digest) with zero remote/S3 traffic.

Implemented in the zlayer bin over the daemon’s DeploymentStorage (SqlxStorage in ZLayer, the ZQL store in ZLayerZQL). The agent itself must NOT depend on the storage layer — that would be a dependency cycle — so the concrete write lives in the bin and is threaded in as this trait object.

record is best-effort: a failed store update is logged by the implementor and MUST NOT fail the deploy/scale that triggered the pull. Breaking the boot-time circular dependency (ZLayer using its own ZataStorage S3 backend as the blob cache) is the whole point — see crate::service::ServiceInstance::set_restore_pin.

Required Methods

fn record<'life0, 'life1, 'life2, 'life3, 'async_trait>( &'life0 self, deployment: &'life1 str, service: &'life2 str, digest: &'life3 str, ) -> Pin<Box<dyn Future<Output = ()> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait, 'life2: 'async_trait, 'life3: 'async_trait,

Persist digest (e.g. "sha256:abc…") as the resolved image digest for service within deployment, keyed by service name in StoredDeployment.resolved_image_digests. Best-effort; never panics.

Dyn Compatibility

This trait is dyn compatible.

In older versions of Rust, dyn compatibility was called "object safety".

Implementors