detached 2175cce

fix(daemon): surface failed session restores at attach

Each of restoreSnapshots' three failure sites (Load, decode, restoreSession)
now also raises a NoticeSnapshotRestore global notice alongside its existing
log line, so a broken saved session is no longer silently dropped at
startup. These run pre-attach, so NotifyGlobal queues into the pending
queue: N failures dedup into one toast with Count N at the first attach,
while each failure still lands in history individually.
commit
2175cce059a1ebce79a1c73e398b2cec70a8295e
author
brice · · 2026-07-21 04:25 UTC
parent
c32a9ec
signature
GPG VerifiedCD88F9E564E0102646FB6077A1A07C7C7C3E349D · arch-gpg Verified for the committer email
2 files · +55 -0
Unified
UnifiedSplit
M internal/usecase/daemon/snapshot.go +6 -0
@@ -69,6 +69,8 @@ func (d *Daemon) restoreSnapshots(ctx context.Context) {
6969 blobs, err := d.snaps.Load()
7070 if err != nil {
7171 d.log.Warn("loading session snapshots failed", "err", err)
72+ d.NotifyGlobal(domain.NoticeError, domain.NoticeSnapshotRestore,
73+ "couldn't load saved sessions after restart", err)
7274 return
7375 }
7476 for _, blob := range blobs {
@@ -80,10 +82,14 @@ func (d *Daemon) restoreSnapshots(ctx context.Context) {
8082 snap, err := snapcodec.Unmarshal(blob.Data)
8183 if err != nil {
8284 d.log.Warn("decoding session snapshot failed", "err", err, "session", blob.Name)
85+ d.NotifyGlobal(domain.NoticeError, domain.NoticeSnapshotRestore,
86+ "couldn't restore session "+blob.Name+" after restart", err)
8387 continue
8488 }
8589 if err := d.restoreSession(ctx, snap); err != nil {
8690 d.log.Warn("restoring session snapshot failed", "err", err, "session", snap.Name)
91+ d.NotifyGlobal(domain.NoticeError, domain.NoticeSnapshotRestore,
92+ "couldn't restore session "+snap.Name+" after restart", err)
8793 }
8894 }
8995 }
@@ -69,6 +69,8 @@ func (d *Daemon) restoreSnapshots(ctx context.Context) {
69 blobs, err := d.snaps.Load()69 blobs, err := d.snaps.Load()
70 if err != nil {70 if err != nil {
71 d.log.Warn("loading session snapshots failed", "err", err)71 d.log.Warn("loading session snapshots failed", "err", err)
72+ d.NotifyGlobal(domain.NoticeError, domain.NoticeSnapshotRestore,
73+ "couldn't load saved sessions after restart", err)
72 return74 return
73 }75 }
74 for _, blob := range blobs {76 for _, blob := range blobs {
@@ -80,10 +82,14 @@ func (d *Daemon) restoreSnapshots(ctx context.Context) {
80 snap, err := snapcodec.Unmarshal(blob.Data)82 snap, err := snapcodec.Unmarshal(blob.Data)
81 if err != nil {83 if err != nil {
82 d.log.Warn("decoding session snapshot failed", "err", err, "session", blob.Name)84 d.log.Warn("decoding session snapshot failed", "err", err, "session", blob.Name)
85+ d.NotifyGlobal(domain.NoticeError, domain.NoticeSnapshotRestore,
86+ "couldn't restore session "+blob.Name+" after restart", err)
83 continue87 continue
84 }88 }
85 if err := d.restoreSession(ctx, snap); err != nil {89 if err := d.restoreSession(ctx, snap); err != nil {
86 d.log.Warn("restoring session snapshot failed", "err", err, "session", snap.Name)90 d.log.Warn("restoring session snapshot failed", "err", err, "session", snap.Name)
91+ d.NotifyGlobal(domain.NoticeError, domain.NoticeSnapshotRestore,
92+ "couldn't restore session "+snap.Name+" after restart", err)
87 }93 }
88 }94 }
89 }95 }
M internal/usecase/daemon/snapshot_restore_test.go +49 -0
@@ -497,6 +497,55 @@ func TestRestoreSnapshotsSkipsCreatedAtOverflow(t *testing.T) {
497497 require.Nil(t, restored)
498498 }
499499
500+// TestRestoreSnapshotsNotifiesGlobalOnUnrestorableSnapshots proves each
501+// unrestorable snapshot is recorded to history individually, while the
502+// pending-global queue dedups them by code into a single toast with the
503+// combined Count. Both blobs decode fine (valid snapcodec bytes, Marshal
504+// itself validates Active/tab range) but fail in restoreSession itself: an
505+// empty embedded session name is rejected deterministically, before any lock
506+// is taken or PTY is opened.
507+func TestRestoreSnapshotsNotifiesGlobalOnUnrestorableSnapshots(t *testing.T) {
508+ badSnap := mustSnapshotBytes(t, snapcodec.Session{Name: ""})
509+ store := &restoreSnapshotStore{blobs: []ports.SnapshotBlob{
510+ {Name: "bad-1", Data: badSnap},
511+ {Name: "bad-2", Data: badSnap},
512+ }}
513+ factory := &restorePTYFactory{}
514+ d := newTestDaemon(t, factory, newNoticeClock())
515+ WithSnapshotStore(store)(d)
516+
517+ // Mirrors real startup (daemon.go's Serve): restoreSnapshots runs off the
518+ // caller goroutine and signals completion by closing d.restoreDone.
519+ d.sessWg.Go(func() { d.restoreSnapshots(d.serveCtx) })
520+
521+ select {
522+ case <-d.restoreDone:
523+ case <-time.After(2 * time.Second):
524+ t.Fatal("restoreSnapshots did not close restoreDone in time")
525+ }
526+
527+ require.Empty(t, factory.opens, "both snapshots must fail before any PTY is opened")
528+ require.Len(t, d.notices.history(), 2, "each failure is recorded individually")
529+ for _, n := range d.notices.history() {
530+ require.Equal(t, domain.NoticeSnapshotRestore, n.Code)
531+ require.Equal(t, domain.NoticeError, n.Severity)
532+ }
533+
534+ tr, _ := newCapturingTransport(t)
535+ sess, ac, err := d.route(ports.Hello{
536+ Version: ports.ProtocolVersion,
537+ Intent: ports.IntentEphemeral,
538+ Size: domain.Size{Cols: 80, Rows: 24},
539+ }, tr)
540+ require.NoError(t, err)
541+ d.firstPaint(sess, ac, ac.size)
542+
543+ toasts := awaitToastCount(t, ac, 1)
544+ require.Equal(t, domain.NoticeSnapshotRestore, toasts[0].Code)
545+ require.Equal(t, 2, toasts[0].Count, "two same-code failures dedup into one toast counted twice")
546+ require.Empty(t, d.notices.drainPending(), "firstPaint must consume the queue")
547+}
548+
500549 func TestNamedRouteWaitsForRestoreBarrier(t *testing.T) {
501550 releaseLoad := make(chan struct{})
502551 loaded := make(chan struct{})
@@ -497,6 +497,55 @@ func TestRestoreSnapshotsSkipsCreatedAtOverflow(t *testing.T) {
497 require.Nil(t, restored)497 require.Nil(t, restored)
498 }498 }
499 499
500+// TestRestoreSnapshotsNotifiesGlobalOnUnrestorableSnapshots proves each
501+// unrestorable snapshot is recorded to history individually, while the
502+// pending-global queue dedups them by code into a single toast with the
503+// combined Count. Both blobs decode fine (valid snapcodec bytes, Marshal
504+// itself validates Active/tab range) but fail in restoreSession itself: an
505+// empty embedded session name is rejected deterministically, before any lock
506+// is taken or PTY is opened.
507+func TestRestoreSnapshotsNotifiesGlobalOnUnrestorableSnapshots(t *testing.T) {
508+ badSnap := mustSnapshotBytes(t, snapcodec.Session{Name: ""})
509+ store := &restoreSnapshotStore{blobs: []ports.SnapshotBlob{
510+ {Name: "bad-1", Data: badSnap},
511+ {Name: "bad-2", Data: badSnap},
512+ }}
513+ factory := &restorePTYFactory{}
514+ d := newTestDaemon(t, factory, newNoticeClock())
515+ WithSnapshotStore(store)(d)
516+
517+ // Mirrors real startup (daemon.go's Serve): restoreSnapshots runs off the
518+ // caller goroutine and signals completion by closing d.restoreDone.
519+ d.sessWg.Go(func() { d.restoreSnapshots(d.serveCtx) })
520+
521+ select {
522+ case <-d.restoreDone:
523+ case <-time.After(2 * time.Second):
524+ t.Fatal("restoreSnapshots did not close restoreDone in time")
525+ }
526+
527+ require.Empty(t, factory.opens, "both snapshots must fail before any PTY is opened")
528+ require.Len(t, d.notices.history(), 2, "each failure is recorded individually")
529+ for _, n := range d.notices.history() {
530+ require.Equal(t, domain.NoticeSnapshotRestore, n.Code)
531+ require.Equal(t, domain.NoticeError, n.Severity)
532+ }
533+
534+ tr, _ := newCapturingTransport(t)
535+ sess, ac, err := d.route(ports.Hello{
536+ Version: ports.ProtocolVersion,
537+ Intent: ports.IntentEphemeral,
538+ Size: domain.Size{Cols: 80, Rows: 24},
539+ }, tr)
540+ require.NoError(t, err)
541+ d.firstPaint(sess, ac, ac.size)
542+
543+ toasts := awaitToastCount(t, ac, 1)
544+ require.Equal(t, domain.NoticeSnapshotRestore, toasts[0].Code)
545+ require.Equal(t, 2, toasts[0].Count, "two same-code failures dedup into one toast counted twice")
546+ require.Empty(t, d.notices.drainPending(), "firstPaint must consume the queue")
547+}
548+
500 func TestNamedRouteWaitsForRestoreBarrier(t *testing.T) {549 func TestNamedRouteWaitsForRestoreBarrier(t *testing.T) {
501 releaseLoad := make(chan struct{})550 releaseLoad := make(chan struct{})
502 loaded := make(chan struct{})551 loaded := make(chan struct{})

Keyboard shortcuts

Single-key shortcuts are on. ctrl k and esc work whatever this says.

ctrl k :
go to a section, branch, tag, file or repository
?
this list
j k
move the row cursor down / up
enter
open the row under the cursor; on Files, a directory opens or closes in place
↑ ↓ ← → home end
in the file tree (focus on a row): previous / next row, close / open a directory or go to its parent / first entry, first / last row. They never open a file
l h
on Files: open a directory or a file / close a directory or go up to its parent, then to the parent page. Elsewhere: into the side panel (on a merge request, its files) / back to the content; there j k move between its controls
g g G
first / last row
/
filter the list; on Files, find a file anywhere in the repository (esc clears, enter opens the best match)
b
switch branch or tag
g s g f g c g b g m g i g r g a
tabs: summary, files, commits, branches & tags, merge requests, issues, releases, actions (when available)
g h
home
w
wide mode: use the whole width of the screen, or not (remembered)
y
copy a permalink pinned to the commit
d u x
diff: split / unified / collapse or expand all
e E
diff: the whole file in view / every file, with the lines the hunks hide
n
merge request: next note in the diff
esc
close a dialog, leave a field

Branches 12

Tags 11

↑↓ moveenter openesc close