agent v0.36.7: isolate shared parent only on create (no peer-group churn)

make-private+make-shared only when first creating the self-bind; re-running it
churns the peer group and orphans the guest's slave (propagation dies).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

CHANGELOG.md

@@ -3,6 +3,13 @@
 All notable changes to **felhom-agent** are recorded here. Update on every code
 change that gets pushed.
 
+## v0.36.7 — isolate the shared parent only on CREATE (no peer-group churn) (2026-06-15)
+
+Follow-up to v0.36.6: make-private+make-shared must run ONLY when the self-bind is first created, not on
+every reconcile — re-doing it churns the peer-group id and ORPHANS the guest`s already-established slave
+(propagation silently dies, guest sees empty). Guarded on the mountpoint check; on a fresh boot it runs
+once before pve-guests so the guest slaves the right group.
+
 ## v0.36.6 — shared parent gets its OWN peer group (make-private first) — ROOT CAUSE of double-bind (2026-06-15)
 
 The shared-parent self-bind INHERITED the root mount`s shared peer group (`/mnt/felhom-drives` was

cmd/felhom-agent/main.go

@@ -44,7 +44,7 @@ import (
 
 // version is the agent version. Overridable at build time with
 // -ldflags "-X main.version=<v>"; defaults to the in-repo CHANGELOG version.
-var version = "0.36.6"
+var version = "0.36.7"
 
 // runGuestHook is the PVE pre-start hook body (`felhom-agent guest-hook <vmid> <phase>`). On the
 // pre-start phase it creates placeholder dirs for any absent bind-mount source so the guest always boots

internal/localapi/intermediary.go

@@ -32,12 +32,15 @@ const sharedParentScript = `#!/bin/sh
 # the shared peer group (slave). Installed + enabled by felhom-agent. Idempotent.
 set -e
 mkdir -p ` + StableParentDir + `
-mountpoint -q ` + StableParentDir + ` || mount --bind ` + StableParentDir + ` ` + StableParentDir + `
+# Isolate + share ONLY when first creating the self-bind (a fresh boot). The self-bind inherits the root
-# make-PRIVATE first: the self-bind inherits the root mount's shared peer group, so without detaching it
+# mount's shared peer group, so make-private detaches it (else binds under it DOUBLE via the root peer),
-# every drive bind under the parent would propagate back (and DOUBLE). Then make-shared = its OWN group,
+# then make-shared gives it its own group whose only slave is the guest's parent bind. Re-running this on
-# whose only slave is the guest's parent bind → binds propagate to the guest exactly once.
+# an existing parent would churn the peer-group id and orphan the guest's slave — so guard on mountpoint.
-mount --make-private ` + StableParentDir + `
+if ! mountpoint -q ` + StableParentDir + `; then
-mount --make-shared ` + StableParentDir + `
+  mount --bind ` + StableParentDir + ` ` + StableParentDir + `
+  mount --make-private ` + StableParentDir + `
+  mount --make-shared ` + StableParentDir + `
+fi
 `
 
 const sharedParentUnitPath = "/etc/systemd/system/felhom-shared-parent.service"
@@ -94,19 +97,22 @@ func (b *GuestBinder) EnsureSharedParent(ctx context.Context) error {
 	if err := b.run(ctx, "mkdir", "-p", StableParentDir); err != nil {
 		return fmt.Errorf("shared-parent: mkdir %s: %w", StableParentDir, err)
 	}
+	// Isolate + share the parent ONLY when first creating the self-bind. The self-bind inherits the root
+	// mount's shared peer group, so make-PRIVATE detaches it (else binds under it double via the root
+	// peer); make-SHARED then gives it its OWN group whose only slave is the guest's parent bind. This
+	// must NOT run on every reconcile: re-doing make-private+make-shared churns the peer-group id and
+	// ORPHANS the guest's already-established slave (propagation silently dies). On a fresh host boot the
+	// parent isn't a mountpoint → this runs once, before pve-guests, so the guest slaves the right group.
 	if !isHostMountpoint(StableParentDir) {
 		if err := b.run(ctx, "mount", "--bind", StableParentDir, StableParentDir); err != nil {
 			return fmt.Errorf("shared-parent: self-bind: %w", err)
 		}
-	}
+		if err := b.run(ctx, "mount", "--make-private", StableParentDir); err != nil {
-	// make-PRIVATE first to detach from the root mount's shared peer group (the self-bind inherits it),
+			return fmt.Errorf("shared-parent: make-private: %w", err)
-	// THEN make-shared so the parent owns its OWN group — otherwise binds under it propagate back via the
+		}
-	// root peer and DOUBLE (observed live as 2 stacked binds per drive).
+		if err := b.run(ctx, "mount", "--make-shared", StableParentDir); err != nil {
-	if err := b.run(ctx, "mount", "--make-private", StableParentDir); err != nil {
+			return fmt.Errorf("shared-parent: make-shared: %w", err)
-		return fmt.Errorf("shared-parent: make-private: %w", err)
+		}
-	}
-	if err := b.run(ctx, "mount", "--make-shared", StableParentDir); err != nil {
-		return fmt.Errorf("shared-parent: make-shared: %w", err)
 	}
 	// Install the boot-persistence unit only when missing OR its content differs from what we ship (so a
 	// unit-template fix deploys) — EnsureSharedParent runs on a periodic reconcile, and re-writing files +