d341fee fix(subagents): skip incompatible journals during discovery (#180) · zut

M docs/subagents.md +4 -2


@@ -90,8 +90,10 @@ prompt that has not received a slot and records its terminal failure durably.
90	90		a cumulative weighted-token budget equal to its resolved model's context window.
91	91		Removed legacy settings, including `tui_subagent_position` and
92	92		`subagents.budget_ratio`, are ignored. Resident journals created before this
93		-	budget policy are incompatible: zut leaves them untouched, reports a recovery
94		-	error, and requires a new child rather than resuming them.
	93	+	budget policy are incompatible: zut leaves them untouched on disk and excludes
	94	+	them from automatic discovery without reporting an error at every startup.
	95	+	They cannot be resumed; start a new child instead. Other journal recovery
	96	+	errors are still reported.
95	97
96	98		Budget accounting charges uncached input, cache writes, and output fully; cache
97	99		reads count at 25%. It is separate from active context-window usage and survives

 a cumulative weighted-token budget equal to its resolved model's context window.
 Removed legacy settings, including `tui_subagent_position` and
 `subagents.budget_ratio`, are ignored. Resident journals created before this
-budget policy are incompatible: zut leaves them untouched, reports a recovery
+budget policy are incompatible: zut leaves them untouched on disk and excludes
-error, and requires a new child rather than resuming them.
+them from automatic discovery without reporting an error at every startup.
+They cannot be resumed; start a new child instead. Other journal recovery
+errors are still reported.
 Budget accounting charges uncached input, cache writes, and output fully; cache
 reads count at 25%. It is separate from active context-window usage and survives

M packages/agent/subagents/resident_manager.go +9 -2


@@ -809,8 +809,9 @@ func (m *ResidentManager) SnapshotFor(childID string) (ResidentSnapshot, bool) {
809	809
810	810		// Reconcile discovers journals left by an earlier host, repairs their bounded
811	811		// projections, and marks queued/running work interrupted. It never creates a
812		-	// live child or replays a prompt. Per-child errors are returned for callers to
813		-	// surface without hiding valid sibling journals.
	812	+	// live child or replays a prompt. Incompatible budget journals are left untouched
	813	+	// and excluded from discovery. Other per-child errors are returned for callers
	814	+	// to surface without hiding valid sibling journals.
814	815		func (m *ResidentManager) Reconcile() []error {
815	816		if m == nil {
816	817		return []error{errors.New("resident manager: unavailable")}

@@ -850,6 +851,12 @@ func (m *ResidentManager) Reconcile() []error {
850	851		}
851	852		childDir := filepath.Join(m.root, childID)
852	853		metadata, spec, reconcileErr := reconcileResidentJournalWithSpec(childDir)
	854	+	// Old budget journals cannot be resumed. Leave them on disk without
	855	+	// reporting the same incompatibility at every host startup. Match the
	856	+	// exact sentinel so a joined journal-close error is still reported.
	857	+	if reconcileErr == ErrIncompatibleResidentBudget {
	858	+	continue
	859	+	}
853	860		if errors.Is(reconcileErr, ErrResidentLeaseBusy) {
854	861		foreign, readErr := ReadResidentMetadata(filepath.Join(childDir, residentMetadataName))
855	862		if readErr != nil && !errors.Is(readErr, os.ErrNotExist) {

 // Reconcile discovers journals left by an earlier host, repairs their bounded
 // projections, and marks queued/running work interrupted. It never creates a
-// live child or replays a prompt. Per-child errors are returned for callers to
+// live child or replays a prompt. Incompatible budget journals are left untouched
-// surface without hiding valid sibling journals.
+// and excluded from discovery. Other per-child errors are returned for callers
+// to surface without hiding valid sibling journals.
 func (m *ResidentManager) Reconcile() []error {
 	if m == nil {
 		return []error{errors.New("resident manager: unavailable")}
 		}
 		childDir := filepath.Join(m.root, childID)
 		metadata, spec, reconcileErr := reconcileResidentJournalWithSpec(childDir)
+		// Old budget journals cannot be resumed. Leave them on disk without
+		// reporting the same incompatibility at every host startup. Match the
+		// exact sentinel so a joined journal-close error is still reported.
+		if reconcileErr == ErrIncompatibleResidentBudget {
+			continue
+		}
 		if errors.Is(reconcileErr, ErrResidentLeaseBusy) {
 			foreign, readErr := ReadResidentMetadata(filepath.Join(childDir, residentMetadataName))
 			if readErr != nil && !errors.Is(readErr, os.ErrNotExist) {

M packages/agent/subagents/resident_manager_test.go +36 -4


@@ -584,7 +584,7 @@ func TestResidentManagerRejectsDuplicateBeforeSecondJournalAcceptance(t *testing
584	584		}
585	585		}
586	586
587		-	func TestResidentManagerReconcileRejectsIncompatibleBudgetJournal(t *testing.T) {
	587	+	func TestResidentManagerReconcileSkipsIncompatibleBudgetJournal(t *testing.T) {
588	588		root := t.TempDir()
589	589		journal, err := OpenResidentJournal(root, "legacy-budget")
590	590		if err != nil {

@@ -611,9 +611,13 @@ func TestResidentManagerReconcileRejectsIncompatibleBudgetJournal(t *testing.T)
611	611		}
612	612		manager := NewResidentManager(root, nil)
613	613		t.Cleanup(func() { _ = manager.Close(context.Background()) })
614		-	errs := manager.Reconcile()
615		-	if len(errs) != 1 \|\| !errors.Is(errs[0], ErrIncompatibleResidentBudget) {
616		-	t.Fatalf("Reconcile errors = %v", errs)
	614	+	for range 2 {
	615	+	if errs := manager.Reconcile(); len(errs) != 0 {
	616	+	t.Fatalf("Reconcile errors = %v", errs)
	617	+	}
	618	+	}
	619	+	if _, err := ReconcileResidentJournal(filepath.Dir(transcript)); !errors.Is(err, ErrIncompatibleResidentBudget) {
	620	+	t.Fatalf("explicit reconciliation error = %v, want incompatible budget", err)
617	621		}
618	622		if snapshots := manager.Snapshot(); len(snapshots) != 0 {
619	623		t.Fatalf("snapshots = %#v", snapshots)

@@ -625,6 +629,34 @@ func TestResidentManagerReconcileRejectsIncompatibleBudgetJournal(t *testing.T)
625	629		if string(unchanged) != legacy {
626	630		t.Fatal("Reconcile modified incompatible journal")
627	631		}
	632	+
	633	+	valid, err := OpenResidentJournal(root, "valid")
	634	+	if err != nil {
	635	+	t.Fatal(err)
	636	+	}
	637	+	spec.ID = "valid"
	638	+	if err := valid.Accept(spec, "keep discoverable"); err != nil {
	639	+	t.Fatal(err)
	640	+	}
	641	+	if err := valid.Close(); err != nil {
	642	+	t.Fatal(err)
	643	+	}
	644	+	corrupt, err := OpenResidentJournal(root, "corrupt")
	645	+	if err != nil {
	646	+	t.Fatal(err)
	647	+	}
	648	+	if err := corrupt.Close(); err != nil {
	649	+	t.Fatal(err)
	650	+	}
	651	+	for range 2 {
	652	+	errs := manager.Reconcile()
	653	+	if len(errs) != 1 \|\| !strings.Contains(errs[0].Error(), "resident child corrupt:") {
	654	+	t.Fatalf("Reconcile errors = %v, want corrupt child error only", errs)
	655	+	}
	656	+	if snapshots := manager.Snapshot(); len(snapshots) != 1 \|\| snapshots[0].ID != "valid" {
	657	+	t.Fatalf("snapshots = %#v, want valid sibling only", snapshots)
	658	+	}
	659	+	}
628	660		}
629	661
630	662		func TestResidentManagerReconcileMakesInterruptedJournalDiscoverableWithoutReplay(t *testing.T) {


@@ -584,7 +584,7 @@ func TestResidentManagerRejectsDuplicateBeforeSecondJournalAcceptance(t *testing
584		}	584		}
585		}	585		}
586			586
587	-	func TestResidentManagerReconcileRejectsIncompatibleBudgetJournal(t *testing.T) {	587	+	func TestResidentManagerReconcileSkipsIncompatibleBudgetJournal(t *testing.T) {
588		root := t.TempDir()	588		root := t.TempDir()
589		journal, err := OpenResidentJournal(root, "legacy-budget")	589		journal, err := OpenResidentJournal(root, "legacy-budget")
590		if err != nil {	590		if err != nil {

@@ -611,9 +611,13 @@ func TestResidentManagerReconcileRejectsIncompatibleBudgetJournal(t *testing.T)
611		}	611		}
612		manager := NewResidentManager(root, nil)	612		manager := NewResidentManager(root, nil)
613		t.Cleanup(func() { _ = manager.Close(context.Background()) })	613		t.Cleanup(func() { _ = manager.Close(context.Background()) })
614	-	errs := manager.Reconcile()	614	+	for range 2 {
615	-	if len(errs) != 1 \|\| !errors.Is(errs[0], ErrIncompatibleResidentBudget) {	615	+	if errs := manager.Reconcile(); len(errs) != 0 {
616	-	t.Fatalf("Reconcile errors = %v", errs)	616	+	t.Fatalf("Reconcile errors = %v", errs)
			617	+	}
			618	+	}
			619	+	if _, err := ReconcileResidentJournal(filepath.Dir(transcript)); !errors.Is(err, ErrIncompatibleResidentBudget) {
			620	+	t.Fatalf("explicit reconciliation error = %v, want incompatible budget", err)
617		}	621		}
618		if snapshots := manager.Snapshot(); len(snapshots) != 0 {	622		if snapshots := manager.Snapshot(); len(snapshots) != 0 {
619		t.Fatalf("snapshots = %#v", snapshots)	623		t.Fatalf("snapshots = %#v", snapshots)

@@ -625,6 +629,34 @@ func TestResidentManagerReconcileRejectsIncompatibleBudgetJournal(t *testing.T)
625		if string(unchanged) != legacy {	629		if string(unchanged) != legacy {
626		t.Fatal("Reconcile modified incompatible journal")	630		t.Fatal("Reconcile modified incompatible journal")
627		}	631		}
			632	+
			633	+	valid, err := OpenResidentJournal(root, "valid")
			634	+	if err != nil {
			635	+	t.Fatal(err)
			636	+	}
			637	+	spec.ID = "valid"
			638	+	if err := valid.Accept(spec, "keep discoverable"); err != nil {
			639	+	t.Fatal(err)
			640	+	}
			641	+	if err := valid.Close(); err != nil {
			642	+	t.Fatal(err)
			643	+	}
			644	+	corrupt, err := OpenResidentJournal(root, "corrupt")
			645	+	if err != nil {
			646	+	t.Fatal(err)
			647	+	}
			648	+	if err := corrupt.Close(); err != nil {
			649	+	t.Fatal(err)
			650	+	}
			651	+	for range 2 {
			652	+	errs := manager.Reconcile()
			653	+	if len(errs) != 1 \|\| !strings.Contains(errs[0].Error(), "resident child corrupt:") {
			654	+	t.Fatalf("Reconcile errors = %v, want corrupt child error only", errs)
			655	+	}
			656	+	if snapshots := manager.Snapshot(); len(snapshots) != 1 \|\| snapshots[0].ID != "valid" {
			657	+	t.Fatalf("snapshots = %#v, want valid sibling only", snapshots)
			658	+	}
			659	+	}
628		}	660		}
629			661
630		func TestResidentManagerReconcileMakesInterruptedJournalDiscoverableWithoutReplay(t *testing.T) {	662		func TestResidentManagerReconcileMakesInterruptedJournalDiscoverableWithoutReplay(t *testing.T) {