feat(#2398): dagify hivectl restart (RestartScoped)

hivectl restart --agent NAME previously composed stop() then start() as
two separate client-side daemon calls glued by CLI-process control flow
— not one DAG, and a dropped hivectl connection mid-restart (ssh drop,
Ctrl-C) left the agent stopped with no automatic follow-up. mara flagged
this as the first target for the 'dagify hivectl commands' issue.

New HostRequest::RestartScoped{scope, graceful} handles it server-side:
each targeted agent now rides exactly one atomic Restart-template DAG
(same one hivectl agents restart / restart-all already use) in the
common non-graceful case. --graceful has no single-DAG template yet, so
it submits the graceful-stop DAGs, awaits them server-side, then submits
the start DAGs — still one daemon call end to end, just not yet a single
DAG (noted as a follow-up). Infra containers restart synchronously as
before (no lease/DAG concept for them).

CLI-side restart() now just makes the one call + waits, same output
shape as before via render_lifecycle.

hive-c0re/src/bin/hivectl.rs

@@ -1642,22 +1642,31 @@ async fn start(socket: &Path, scope: hive_host_sock::LifecycleScope, no_wait: bo
     rendered
 }
 
-/// Restart = `stop` then `start` over the same scope, composed client-side
-/// from the two daemon ops (no dedicated wire op). The stop phase honours
-/// `--graceful`; if it reports a failure (`stop` returns `Err`) the `?`
-/// short-circuits before the start phase, so a half-stopped hive isn't
-/// blindly started over — the operator sees the stop errors and can recover.
+/// Restart — one `RestartScoped` daemon call, server-side DAG-based (see
+/// issue tracker "dagify hivectl commands"). Each targeted agent rides one
+/// atomic `Restart` DAG (or, with `--graceful`, a server-awaited
+/// graceful-stop DAG followed by a start DAG); infra containers restart
+/// synchronously. Unlike the old client-side stop-then-start compose, a
+/// dropped `hivectl` connection mid-restart no longer leaves an agent
+/// stopped with no automatic follow-up — the daemon owns the whole
+/// sequence once this call is made.
 ///
-/// No `--no-wait` here on purpose: the stop DAGs must complete before
-/// the start submits, otherwise the start's `wanted = Up` write would
-/// land before the queued stops execute and turn them into noops.
+/// No `--no-wait` here on purpose, same as before: the operator wants to
+/// see the restart actually land, not just get queued.
 async fn restart(
     socket: &Path,
     scope: hive_host_sock::LifecycleScope,
     graceful: bool,
 ) -> Result<()> {
-    stop(socket, scope.clone(), graceful, false).await?;
-    start(socket, scope, false).await
+    let resp = hive_c0re::client::request(
+        socket,
+        hive_host_sock::HostRequest::RestartScoped { scope, graceful },
+    )
+    .await
+    .with_context(|| format!("connect to daemon socket {}", socket.display()))?;
+    let rendered = render_lifecycle(&resp, "restart queued");
+    wait_for_dags(socket, resp.queued_dags.unwrap_or_default(), false).await?;
+    rendered
 }
 
 /// A [`LifecycleScope`](hive_host_sock::LifecycleScope) targeting exactly one

hive-c0re/src/server.rs

@@ -94,6 +94,9 @@ async fn dispatch(req: &HostRequest, coord: Arc<Coordinator>) -> HostResponse {
             HostRequest::Kill { name } => submit_single(&coord, name, Verb::Kill),
             HostRequest::Restart { name } => submit_single(&coord, name, Verb::Restart),
             HostRequest::RestartAll => handle_restart_all(&coord).await?,
+            HostRequest::RestartScoped { scope, graceful } => {
+                handle_restart_scoped(&coord, scope, *graceful).await?
+            }
             HostRequest::Stop { scope, graceful } => {
                 // Resolve the scope to explicit container names at the entry
                 // point, then operate on names — never pass the bare "all
@@ -631,6 +634,85 @@ async fn handle_start(
     Ok(resp)
 }
 
+/// Restart containers hive-wide (`hivectl restart`) — the DAG-based
+/// sibling of [`handle_stop`]/[`handle_start`], replacing the old
+/// client-side stop-then-start composition (issue tracker "dagify hivectl
+/// commands"). Each targeted agent gets exactly one atomic `Restart`
+/// template DAG in the common (non-graceful) case, so it survives a
+/// dropped `hivectl` connection the same way `handle_restart_all` already
+/// does. `graceful` has no single-DAG template yet, so it falls back to
+/// submitting the graceful-stop DAGs, waiting them out server-side, then
+/// submitting the start DAGs — still one daemon call end to end, unlike
+/// the old CLI-side compose. Infra containers have no lease/DAG and
+/// restart synchronously (stop then start).
+async fn handle_restart_scoped(
+    coord: &Arc<Coordinator>,
+    scope: &LifecycleScope,
+    graceful: bool,
+) -> Result<HostResponse> {
+    tracing::info!(?scope, graceful, "restart");
+    let agents = scoped_agents(scope).await?;
+    let infra = scoped_infra(scope);
+    let mut ok_items: Vec<String> = Vec::new();
+    let mut errors: Vec<String> = Vec::new();
+    let mut queued: Vec<u64> = Vec::new();
+
+    if graceful {
+        let mut stop_ids = Vec::new();
+        for agent in &agents {
+            stop_ids.push(crate::job_queue::submit::graceful_stop(
+                coord,
+                agent,
+                crate::job_queue::Source::Manual,
+                "manual via hivectl restart --graceful".to_owned(),
+            ));
+        }
+        // Graceful drains can take a while; bound the wait generously —
+        // same shape as `handle_stop`'s hard-stop-before-infra wait, just
+        // a longer ceiling since drains are minutes, not seconds.
+        await_dags(coord, &stop_ids, std::time::Duration::from_mins(10)).await;
+        for agent in &agents {
+            queued.push(crate::job_queue::submit::start(
+                coord,
+                agent,
+                crate::job_queue::Source::Manual,
+                "manual via hivectl restart --graceful".to_owned(),
+            ));
+            ok_items.push(agent.clone());
+        }
+    } else {
+        for agent in &agents {
+            queued.push(crate::job_queue::submit::restart(
+                coord,
+                agent,
+                crate::job_queue::Source::Manual,
+                "manual restart via hivectl restart".to_owned(),
+            ));
+            ok_items.push(agent.clone());
+        }
+    }
+
+    for &container in &infra {
+        let name = container.unit_name();
+        let res = async {
+            crate::priv_client::control_infra_container(container, InfraAction::Stop).await?;
+            crate::priv_client::control_infra_container(container, InfraAction::Start).await
+        }
+        .await;
+        match res {
+            Ok(()) => ok_items.push(name.to_owned()),
+            Err(e) => {
+                tracing::warn!(%name, error = ?e, "restart: infra restart failed");
+                errors.push(format!("{name}: {e:#}"));
+            }
+        }
+    }
+
+    let mut resp = finish_lifecycle(ok_items, &errors);
+    resp.queued_dags = Some(queued);
+    Ok(resp)
+}
+
 /// Resolve which sub-agent logical names a scope targets: every live
 /// container (from `lifecycle::list`) when `agents` is set or the scope is
 /// "everything", plus any explicit `agent_names`. Returns de-duplicated

hive-host-sock/src/lib.rs

@@ -39,6 +39,23 @@ pub enum HostRequest {
     /// wrapper for `hivectl agents restart-all`; iterates the live
     /// container list and restarts each one.
     RestartAll,
+    /// Restart containers hive-wide (`hivectl restart`), scoped like
+    /// `Stop`/`Start`. Each targeted agent rides exactly one DAG server-side
+    /// (the `Restart` template, hard stop + reconcile — or, when `graceful`
+    /// is set, a submitted graceful-stop DAG the server itself waits out
+    /// before submitting the start DAG) rather than the old client-side
+    /// stop-then-start composition: a dropped `hivectl` connection mid-way
+    /// used to leave the agent stopped with no automatic follow-up, since
+    /// nothing durable remembered "finish the restart" once the CLI process
+    /// was gone. Infra containers have no lease/DAG and restart
+    /// synchronously (stop then start), same as before. Scope semantics
+    /// match `Stop`/`Start` (all-false = everything).
+    RestartScoped {
+        #[serde(default)]
+        scope: LifecycleScope,
+        #[serde(default)]
+        graceful: bool,
+    },
     /// Apply pending config to a managed container.
     Rebuild { name: String },
     /// List managed containers.