| 1 | package activator |
| 2 | |
| 3 | import ( |
| 4 | "context" |
| 5 | "errors" |
| 6 | "github.com/carlosframework/platform/internal/hostclaim" |
| 7 | "github.com/carlosframework/platform/internal/hostwrite" |
| 8 | "github.com/carlosframework/platform/internal/moves" |
| 9 | "github.com/carlosframework/platform/internal/provision" |
| 10 | "github.com/carlosframework/platform/internal/registry" |
| 11 | "os/exec" |
| 12 | "testing" |
| 13 | "time" |
| 14 | ) |
| 15 | |
| 16 | func TestStartAdmittedPreparedInputsAndRestoreOutsideProbeWindow(t *testing.T) { |
| 17 | manifest := &moves.InstallManifest{Op: "op", Host: "x.example", SnapshotVersion: "pinned"} |
| 18 | inputs := &provision.MoveInputs{Release: "release", OverlayETag: "overlay"} |
| 19 | policy := startPolicy{MaxStarts: 5, Window: time.Millisecond, Deadline: time.Now().Add(time.Minute), Probe: true, Prepared: manifest, Inputs: inputs} |
| 20 | obs := hostclaim.Observed{Host: "x.example", ETag: "one", Claim: &hostclaim.Claim{Holder: &hostclaim.Holder{Location: "instances/x.example/inc", Session: 1, Generation: "one"}, Op: hostclaim.Op{Kind: "landing", ID: "op", Phase: "probing", Exec: "box"}}} |
| 21 | work, _ := (hostwrite.Config{}).Select(t.Context(), obs.Host, "spawn") |
| 22 | stop := errors.New("stop before kernel start") |
| 23 | restored, materialized, journaled := false, false, false |
| 24 | a := &Service{cfg: Config{HostWrites: hostwrite.Config{Enabled: true}, Claims: &ClaimRuntime{Box: "box", Read: func(context.Context, string) (hostclaim.Observed, error) { return obs, nil }, Prepare: func(ctx context.Context, rt registry.Route, got hostclaim.Observed, p startPolicy) (ClaimStart, error) { |
| 25 | if p.Prepared != manifest || p.Inputs != inputs { |
| 26 | t.Fatal("C1 inputs lost before preparation") |
| 27 | } |
| 28 | return ClaimStart{RoleLauncher: &RoleLaunch{}, CurrentAttempt: func() hostclaim.WakeRetryAttempt { return hostclaim.WakeRetryAttempt{} }, Host: obs.Host, Work: work, Identity: hostwrite.Identity{Config: "c", Release: "r"}, |
| 29 | OpenProbe: func(context.Context, time.Time) (time.Time, error) { journaled = true; return time.Now(), nil }, |
| 30 | Restore: func(ctx context.Context) error { |
| 31 | restored = true |
| 32 | deadline, ok := ctx.Deadline() |
| 33 | if !ok || !deadline.Equal(policy.Deadline) { |
| 34 | t.Fatal("restore consumed probe window") |
| 35 | } |
| 36 | time.Sleep(5 * policy.Window) |
| 37 | return ctx.Err() |
| 38 | }, |
| 39 | Materialize: func(context.Context) error { materialized = true; return stop }, Build: func() *exec.Cmd { return nil }, Ready: func(context.Context, *exec.Cmd) error { return nil }, |
| 40 | Reserve: func(context.Context, int) error { return nil }, Failed: func(context.Context, int) error { return nil }, Healthy: func(context.Context, int, *exec.Cmd) error { return nil }, Replicate: func(context.Context, string) (Rep, error) { return nil, nil }, Publish: func(context.Context, hostclaim.Observed) (hostclaim.Observed, error) { return obs, nil }, Retain: func(Proc, Rep, error) {}, |
| 41 | }, nil |
| 42 | }}}} |
| 43 | err := a.startAdmitted(t.Context(), &actEntry{host: obs.Host}, obs, policy) |
| 44 | if !errors.Is(err, stop) || !restored || !materialized || journaled { |
| 45 | t.Fatalf("err=%v restore=%v materialize=%v journal=%v", err, restored, materialized, journaled) |
| 46 | } |
| 47 | policy.Deadline = time.Now().Add(-time.Second) |
| 48 | if err := a.startAdmitted(t.Context(), &actEntry{host: obs.Host}, obs, policy); !errors.Is(err, ErrAttemptOver) { |
| 49 | t.Fatalf("cut deadline: %v", err) |
| 50 | } |
| 51 | } |
| 52 | |
| 53 | func TestClaimProbeWindowJournalsFirstStartAndSurvivesRetry(t *testing.T) { |
| 54 | now := time.Unix(1000, 0) |
| 55 | var saved time.Time |
| 56 | policy := startPolicy{Probe: true, Window: 30 * time.Second, Deadline: now.Add(time.Minute)} |
| 57 | journal := func(_ context.Context, proposed time.Time) (time.Time, error) { |
| 58 | if saved.IsZero() { |
| 59 | saved = proposed |
| 60 | } |
| 61 | return saved, nil |
| 62 | } |
| 63 | b := &probeWindow{policy: policy, now: func() time.Time { return now }} |
| 64 | // Restore took longer than the window, but has not started the app. |
| 65 | now = now.Add(40 * time.Second) |
| 66 | if err := b.start(t.Context(), journal); err != nil { |
| 67 | t.Fatal(err) |
| 68 | } |
| 69 | first := saved |
| 70 | now = now.Add(time.Second) |
| 71 | if err := b.start(t.Context(), journal); err != nil || saved != first { |
| 72 | t.Fatal("retry reset", err, saved) |
| 73 | } |
| 74 | b = &probeWindow{policy: policy, now: func() time.Time { return now }} |
| 75 | if err := b.start(t.Context(), journal); err != nil || b.opened != first { |
| 76 | t.Fatal("restart reset", err, b.opened) |
| 77 | } |
| 78 | now = policy.Deadline |
| 79 | if !errors.Is(b.result(nil), ErrAttemptOver) { |
| 80 | t.Fatal("late readiness accepted after cut") |
| 81 | } |
| 82 | b.policy.Deadline = time.Time{} |
| 83 | now = first.Add(policy.Window) |
| 84 | if err := b.result(nil); !errors.Is(err, ErrAttemptOver) { |
| 85 | t.Fatalf("late readiness ignored context: %v", err) |
| 86 | } |
| 87 | if err := b.start(t.Context(), journal); !errors.Is(err, ErrAttemptOver) { |
| 88 | t.Fatalf("expired retry: %v", err) |
| 89 | } |
| 90 | } |
| 91 | |
| 92 | // Ported from the model's TestProbeWindow and wProbeWindowAt (R12-2, |
| 93 | // INV-19, internal/sim/hostlife/witness_test.go): readiness k seconds after |
| 94 | // the attempt's first start. Main's boundary is adopted (ruling H17): the |
| 95 | // window and the cut deadline both fail at the instant, not after it |
| 96 | // (claimwindow.go: !now.Before), so 29 is accepted, 30 and 31 are not. |
| 97 | // |
| 98 | // The loop around probeWindow is not reachable here. Its Ready hook runs |
| 99 | // only after RoleLaunch.Start, which needs the isolated root, so the |
| 100 | // readiness decision is tested where the loop reads it: probeWindow.start |
| 101 | // before a start and probeWindow.result after readiness. |
| 102 | func TestProbeWindowBoundaries(t *testing.T) { |
| 103 | t0 := time.Date(2026, 10, 6, 9, 0, 0, 0, time.UTC) |
| 104 | far := t0.Add(time.Hour) |
| 105 | newWindow := func(deadline time.Time, now *time.Time) (*probeWindow, func(context.Context, time.Time) (time.Time, error), *int) { |
| 106 | var saved time.Time |
| 107 | calls := 0 |
| 108 | journal := func(_ context.Context, proposed time.Time) (time.Time, error) { |
| 109 | calls++ |
| 110 | if saved.IsZero() { |
| 111 | saved = proposed |
| 112 | } |
| 113 | return saved, nil |
| 114 | } |
| 115 | return &probeWindow{policy: startPolicy{MaxStarts: 5, Probe: true, Window: 30 * time.Second, Deadline: deadline}, now: func() time.Time { return *now }}, journal, &calls |
| 116 | } |
| 117 | for k, want := range map[int]bool{29: true, 30: false, 31: false} { |
| 118 | now := t0 |
| 119 | b, journal, _ := newWindow(far, &now) |
| 120 | if err := b.start(t.Context(), journal); err != nil { |
| 121 | t.Fatalf("k=%d first start: %v", k, err) |
| 122 | } |
| 123 | now = t0.Add(time.Duration(k) * time.Second) |
| 124 | err := b.result(nil) |
| 125 | if (err == nil) != want || (err != nil && !errors.Is(err, ErrAttemptOver)) { |
| 126 | t.Fatalf("k=%d: readiness result %v, accepted want %v", k, err, want) |
| 127 | } |
| 128 | } |
| 129 | // The environment becomes ready only after the window: no further start |
| 130 | // is journaled, and the first start's time is not reset. |
| 131 | now := t0 |
| 132 | b, journal, calls := newWindow(far, &now) |
| 133 | if err := b.start(t.Context(), journal); err != nil || *calls != 1 { |
| 134 | t.Fatalf("first start: %v (journal calls %d)", err, *calls) |
| 135 | } |
| 136 | now = t0.Add(30 * time.Second) |
| 137 | if err := b.start(t.Context(), journal); !errors.Is(err, ErrAttemptOver) || *calls != 1 { |
| 138 | t.Fatalf("start after the window: %v (journal calls %d)", err, *calls) |
| 139 | } |
| 140 | // The absolute cut deadline bounds the whole, inside the window. |
| 141 | now = t0 |
| 142 | c, journal, _ := newWindow(t0.Add(4*time.Second), &now) |
| 143 | if err := c.start(t.Context(), journal); err != nil { |
| 144 | t.Fatal(err) |
| 145 | } |
| 146 | now = t0.Add(3 * time.Second) |
| 147 | if err := c.result(nil); err != nil { |
| 148 | t.Fatalf("readiness before the cut deadline: %v", err) |
| 149 | } |
| 150 | now = t0.Add(4 * time.Second) |
| 151 | if err := c.result(nil); !errors.Is(err, ErrAttemptOver) { |
| 152 | t.Fatalf("readiness at the cut deadline: %v", err) |
| 153 | } |
| 154 | now = t0.Add(5 * time.Second) |
| 155 | if err := c.start(t.Context(), journal); !errors.Is(err, ErrAttemptOver) { |
| 156 | t.Fatalf("start past the cut deadline: %v", err) |
| 157 | } |
| 158 | // A readiness already accepted is not revisited by a later clock. |
| 159 | now = t0 |
| 160 | d, journal, _ := newWindow(far, &now) |
| 161 | if err := d.start(t.Context(), journal); err != nil { |
| 162 | t.Fatal(err) |
| 163 | } |
| 164 | d.ready = true |
| 165 | now = t0.Add(time.Hour) |
| 166 | if err := d.result(nil); err != nil { |
| 167 | t.Fatalf("accepted readiness revisited: %v", err) |
| 168 | } |
| 169 | // Five starts at most, whatever the clock says. |
| 170 | a := &Service{cfg: Config{HostWrites: hostwrite.Config{Enabled: true}, Claims: &ClaimRuntime{Box: "box", Read: func(context.Context, string) (hostclaim.Observed, error) { return hostclaim.Observed{}, nil }, Prepare: func(context.Context, registry.Route, hostclaim.Observed, startPolicy) (ClaimStart, error) { |
| 171 | t.Fatal("six starts were allowed to prepare") |
| 172 | return ClaimStart{}, nil |
| 173 | }}}} |
| 174 | obs := hostclaim.Observed{Host: "x.example"} |
| 175 | if err := a.startAdmitted(t.Context(), &actEntry{host: obs.Host}, obs, startPolicy{MaxStarts: 6, Window: 30 * time.Second, Probe: true}); !errors.Is(err, hostwrite.ErrUnavailable) { |
| 176 | t.Fatalf("a sixth start: %v", err) |
| 177 | } |
| 178 | } |
| 179 | |
amadan