astrid-runtime/astrid · error · GatewayError::Internal

daemon request: {e}

Error message

daemon request: {e}

What it means

The gateway's `daemon_internal` helper wraps any `anyhow::Error` raised while proxying a request to the astrid daemon into a 500 `GatewayError::Internal` with the message "daemon request: {e}". It is the non-typed catch-all for daemon round-trips; unlike the bus-direct kernel path (`daemon_kernel_error`), it cannot distinguish a timeout (504) and deliberately flattens every daemon-side failure into 500. This migration-to-typed-errors gap is acknowledged in the source comments as pending follow-up work.

Source

Thrown at crates/astrid-gateway/src/routes/principals.rs:603

    req.extensions()
        .get::<CallerContext>()
        .ok_or(GatewayError::Unauthorized)
}

// Both helpers consume their argument logically (wrap and discard);
// clippy::needless_pass_by_value fires because they only `Display` /
// `Debug` the value. Taking by value keeps `map_err(daemon_internal)`
// usable as a one-line closure replacement throughout the routes —
// the by-reference shape would force every call site to write
// `map_err(|e| daemon_internal(&e))`.
// The admin-client request path still surfaces `anyhow::Error` (it is not part
// of this change's typed-error migration — a follow-up); every failure here maps
// to 500. The bus-direct kernel-request path uses the typed
// [`daemon_kernel_error`](crate::routes::daemon_kernel_error) instead, which can
// distinguish a 504 timeout.
#[allow(clippy::needless_pass_by_value)]
pub(crate) fn daemon_internal(e: anyhow::Error) -> GatewayError {
    GatewayError::Internal(anyhow::anyhow!("daemon request: {e}"))
}

#[allow(clippy::needless_pass_by_value)]
pub(crate) fn unexpected(other: AdminResponseBody) -> GatewayError {
    GatewayError::Internal(anyhow::anyhow!(
        "unexpected admin response shape: {other:?}"
    ))
}

/// Read the request body as JSON, capping at 64 `KiB` to bound any
/// pathological inbound on the otherwise-unauthenticated edge.
pub(crate) async fn read_json_body<T: serde::de::DeserializeOwned>(
    req: Request<axum::body::Body>,
) -> GatewayResult<T> {
    let bytes = axum::body::to_bytes(req.into_body(), 64 * 1024)
        .await
        .map_err(|e| GatewayError::BadRequest(format!("body read: {e}")))?;
    Ok(serde_json::from_slice(&bytes)?)

View on GitHub (pinned to affd8760f4)

Solutions

1. Check the daemon is running and reachable (process status, socket/endpoint config) and restart it if needed.
2. Inspect the inner `{e}` text in the 500 body/logs — it carries the underlying transport/decode error; fix that root cause.
3. If the failure is a timeout, migrate this call path to `daemon_kernel_error` so clients get a proper 504 instead of 500.
4. Verify gateway and daemon versions match so request/response shapes deserialize correctly.

Example fix

// before
resp.map_err(daemon_internal)?
// after
resp.await.map_err(daemon_kernel_error)? // typed mapping: timeouts become 504, kernel errors become 4xx

// health-check the daemon before issuing the request
let healthy = reqwest::get("http://gateway/healthz").await?.status().is_success();
if !healthy { return Err("daemon backend unavailable".into()); }

match gateway.call(admin_req).await {
    Ok(resp) => resp,
    Err(e) if e.to_string().starts_with("daemon request:") => {
        // retry once, then fail over / alert
        gateway.call(admin_req).await.map_err(report_and_fallback)?
    }
    Err(e) => return Err(e),
}

When it happens

Trigger: Any admin/principal route handler that awaits a daemon request and receives an Err (transport failure, daemon unavailable, daemon returned an error response, deserialization failure) converts it via `daemon_internal(e)`.

Common situations: The astrid daemon process is down or restarting while the gateway stays up; wrong daemon socket/endpoint configuration; daemon overloaded so the request times out (surfaced as 500 instead of 504 due to the untyped mapping); version mismatch so the daemon's response fails to deserialize.

Related errors

• unexpected response shape for GetCapsuleMetadata: {other:?}
• unexpected admin response shape: {other:?}
• daemon request: {e}
• daemon kernel-request: {e}
• gateway has no live capsule provider probe

AI-assisted analysis of astrid-runtime/astrid@affd8760f4 (2026-09-09). Data as JSON: /api/errors/31f31e1d01dd4f65. Report an issue: GitHub.