From 0a9832768500e6a4b86f90a0011e0f9ad84d63ee Mon Sep 17 00:00:00 2001
From: damocles <damocles@hyperhive>
Date: Sun, 31 May 2026 11:54:33 +0200
Subject: [PATCH] lifecycle: only prebuild when there's downtime to shave (mara
 on #721)

---
 hive-c0re/src/lifecycle.rs | 36 +++++++++++++++++++-----------------
 1 file changed, 19 insertions(+), 17 deletions(-)

diff --git a/hive-c0re/src/lifecycle.rs b/hive-c0re/src/lifecycle.rs
index 8c9040fb..1f59dcc9 100644
--- a/hive-c0re/src/lifecycle.rs
+++ b/hive-c0re/src/lifecycle.rs
@@ -425,22 +425,25 @@ pub async fn rebuild_no_meta(
         // the new profile and the new EXTRA_NSPAWN_FLAGS in one go.
         // If the container was already stopped, `update` builds + sets
         // the profile and we leave it stopped.
-        //
-        // Pre-build the system toplevel **before** stopping the
-        // container so the container is only down for the
-        // profile-swap + restart, not for the full nix evaluation +
-        // fetch + build cycle (#706). Failure here aborts before we
-        // touch the running container — the agent keeps serving its
-        // previous generation while the operator looks at the eval
-        // error. `nixos-container update` then finds the toplevel
-        // already in the store and skips straight to the profile
-        // swap.
         let was_running = is_running(name).await;
-        prebuild_toplevel(name, &flake_ref).await?;
         set_nspawn_flags(&container, agent_dir, claude_dir, notes_dir)?;
         set_resource_limits(&container)?;
         systemd_daemon_reload().await?;
         if was_running {
+            // Pre-build the system toplevel **before** stopping the
+            // running container so the agent keeps serving its
+            // previous generation while the eval + fetch + build
+            // happens out-of-band (#706). `nixos-container update`
+            // then finds the toplevel cached and skips straight to
+            // the profile-swap + restart — downtime collapses to that
+            // window only. Build failures surface here, before we
+            // touch the container.
+            //
+            // When the container is already stopped there's no
+            // downtime to shave (mara on #721#9007) — let `update`
+            // do the build inline rather than evaluating the flake
+            // twice for nothing.
+            prebuild_toplevel(name, &flake_ref).await?;
             run(&["stop", &container]).await?;
         }
         run(&["update", &container, "--flake", &flake_ref]).await?;
@@ -450,12 +453,11 @@ pub async fn rebuild_no_meta(
             Ok(())
         }
     } else {
-        // First spawn: pre-build for parity (`nixos-container create`
-        // builds + creates the container, so warming the store first
-        // doesn't shave downtime — there's none to shave — but it
-        // surfaces eval / fetch errors before `create` half-spawns a
-        // container record that the operator then has to clean up).
-        prebuild_toplevel(name, &flake_ref).await?;
+        // First spawn: no running container, no downtime to shave.
+        // `nixos-container create` builds + creates atomically — if
+        // the build fails, no container record is left around to
+        // clean up — so a pre-build adds nothing but a duplicate
+        // eval (mara on #721#9007).
         run(&["create", &container, "--flake", &flake_ref]).await?;
         set_nspawn_flags(&container, agent_dir, claude_dir, notes_dir)?;
         set_resource_limits(&container)?;