Merge branch 'feat/r4g-view' into feat/r4g

This commit is contained in:
2026-10-23 16:20:17 +02:00
14 changed files with 515 additions and 113 deletions
@@ -602,3 +602,97 @@ func TestProcessWholeWritesTheAnswerToAnAskNeedingNoProof(t *testing.T) { // 129
}
wait(func(s view.Snapshot) bool { return len(s.Asks) == 1 && s.Asks[0].Key == mkey }, "the answered ask never left the list")
}
// TestProcessWholeShowsTheBacklogOfBuildAsks starts the program whole against a bus holding the go seat's work-queue stream
// with its three shared workers, three check asks waiting and one held by a node: the socket's backlog row names three
// waiting, one held, and the oldest's age (1887/A8a AC; compiled in the unit run, run by the merge check).
func TestProcessWholeShowsTheBacklogOfBuildAsks(t *testing.T) { // 1887/A8a AC
srv, err := server.NewServer(&server.Options{Host: "127.0.0.1", Port: -1, JetStream: true, StoreDir: t.TempDir(), NoLog: true, NoSigs: true})
if err != nil {
t.Fatal(err)
}
go srv.Start()
if !srv.ReadyForConnections(10 * time.Second) {
t.Fatal("the bus did not come up")
}
t.Cleanup(srv.Shutdown)
nc, err := nats.Connect(srv.ClientURL())
if err != nil {
t.Fatal(err)
}
t.Cleanup(nc.Close)
js, err := jetstream.New(nc)
if err != nil {
t.Fatal(err)
}
ctx, cancel := context.WithTimeout(context.Background(), 60*time.Second)
defer cancel()
stream, _ := delivery.AskStream(delivery.AskGo)
st, err := js.CreateStream(ctx, jetstream.StreamConfig{Name: stream, Subjects: []string{"mesh.seat.node-build-go.accept.>"}, Retention: jetstream.WorkQueuePolicy})
if err != nil {
t.Fatal(err)
}
var checkWorker jetstream.Consumer
for _, kind := range delivery.AskKinds {
name, _ := delivery.AskWorker(delivery.AskGo, kind)
subject, _ := delivery.AskSubject(delivery.AskGo, kind)
c, err := st.CreateOrUpdateConsumer(ctx, jetstream.ConsumerConfig{Durable: name, FilterSubject: subject, AckPolicy: jetstream.AckExplicitPolicy})
if err != nil {
t.Fatal(err)
}
if kind == delivery.AskCheck {
checkWorker = c
}
}
subject, _ := delivery.AskSubject(delivery.AskGo, delivery.AskCheck)
oldest := time.Now().UTC().Add(-2 * time.Minute)
for i, id := range []string{"m0000001aaaaaa", "m0000002bbbbbb", "m0000003cccccc", "m0000004dddddd"} {
body, err := delivery.EncodeStepAsk(delivery.StepAsk{ID: id, Kind: delivery.AskCheck, Language: delivery.AskGo, Asker: "mesh-delivery-checker", At: oldest.Add(time.Duration(i) * time.Second),
Check: &delivery.CheckJob{Fetch: map[string]any{"repository": "novox/x", "commit": "0123456789abcdef0123456789abcdef01234567"}, Levels: []string{delivery.LevelUnit}}})
if err != nil {
t.Fatal(err)
}
if _, err := js.Publish(ctx, subject, body); err != nil {
t.Fatal(err)
}
}
// A node pulls one and holds it unacknowledged.
held, err := checkWorker.Fetch(1)
if err != nil {
t.Fatal(err)
}
for range held.Messages() {
}
dir, err := os.MkdirTemp(os.Getenv("TMPDIR"), "i")
if err != nil {
t.Fatal(err)
}
t.Cleanup(func() { os.RemoveAll(dir) })
broker := filepath.Join(dir, "broker")
cred, _ := json.Marshal(credential{URL: srv.ClientURL(), Node: "n", Module: "mesh-delivery-view", User: "mesh-delivery-view", Password: "x"})
if err := os.WriteFile(broker, cred, 0o600); err != nil {
t.Fatal(err)
}
socket := filepath.Join(dir, "run", "cli.sock")
env := map[string]string{"MESH_BROKER_FILE": broker, "MESH_DELIVERY_SOCKET": socket, "MESH_DELIVERY_WEB": "127.0.0.1:0"}
runCtx, stop := context.WithCancel(ctx)
done := make(chan error, 1)
go func() { done <- run(runCtx, func(k string) string { return env[k] }, io.Discard) }()
t.Cleanup(func() { stop(); <-done })
var last view.Snapshot
for deadline := time.Now().Add(20 * time.Second); time.Now().Before(deadline); time.Sleep(100 * time.Millisecond) {
s, err := view.Ask(socket)
if err != nil {
continue
}
last = s
for _, b := range s.Backlog {
if b.Language == "go" && b.Kind == "check" && b.Waiting == 3 && b.Held == 1 && b.OldestSecs >= 118 {
return
}
}
}
t.Fatalf("the backlog never showed three waiting asks with the oldest's age; last: %+v", last.Backlog)
}
@@ -138,6 +138,7 @@ func run(ctx context.Context, getenv func(string) string, logOut io.Writer) erro
return err
}
model.SetWriter(kvWriter{ctx: ctx, js: js})
model.SetBacklog(view.BacklogOver(js, 5*time.Second))
for _, bucket := range []string{delivery.HaveBucket, view.AssignmentsBucket, view.BuildsBucket, delivery.PresenceBucket, delivery.StuckBucket, delivery.CandidatesBucket, delivery.ModulesBucket, delivery.AsksBucket, delivery.AnswersBucket, delivery.BuilderBuildsBucket, delivery.AssignerDifferencesBucket, delivery.AssignerAsksBucket, delivery.ViewAnswersBucket} {
go follow(ctx, js, bucket, model, logger)
}
@@ -0,0 +1,116 @@
package view
import (
"context"
"fmt"
"time"
"github.com/nats-io/nats.go/jetstream"
"git.novox.be/novox/mesh-sdk/go/delivery"
)
// The backlog of the build seats (novox/hq ADR 0525 §4, issue 1887 A8a): for each language and kind of ask, how many asks wait
// on the seat's stream for a node to pull them, how many a node holds without having finished them, and how old the oldest
// waiting ask is. It is read from the stream's shared pull worker (delivery.AskWorker) when the view is asked: NumPending
// waits, NumAckPending is held by a node, and the oldest waiting ask is the stream's message at the worker's next sequence.
// BacklogRow is one language's kind of ask: Waiting asks not yet pulled, Held asks pulled by a node and not yet ended, and the
// oldest waiting ask's time and age in seconds. Error says why the row could not be read, never dropping it silently.
type BacklogRow struct {
Language string `json:"language"`
Kind string `json:"kind"`
Waiting uint64 `json:"waiting"`
Held uint64 `json:"held"`
Oldest time.Time `json:"oldest,omitzero"`
OldestSecs int64 `json:"oldest-secs,omitempty"`
Error string `json:"error,omitempty"`
}
// Backlog reads the seats' workers. Info is the worker's counts and the stream sequence it delivers next; Message is the body
// of the first message of the stream at or after seq on the subject.
type Backlog struct {
Info func(stream, worker string) (waiting, held, next uint64, err error)
Message func(stream, subject string, seq uint64) ([]byte, error)
Now func() time.Time
}
// Rows are the rows of every language and kind that has an ask waiting or held, or that could not be read.
func (b *Backlog) Rows() []BacklogRow {
now := time.Now
if b.Now != nil {
now = b.Now
}
out := []BacklogRow{}
for _, lang := range delivery.AskLanguages {
stream, _ := delivery.AskStream(lang)
for _, kind := range delivery.AskKinds {
worker, _ := delivery.AskWorker(lang, kind)
subject, _ := delivery.AskSubject(lang, kind)
row := BacklogRow{Language: lang, Kind: kind}
waiting, held, next, err := b.Info(stream, worker)
if err != nil {
row.Error = err.Error()
out = append(out, row)
continue
}
row.Waiting, row.Held = waiting, held
if waiting > 0 {
body, err := b.Message(stream, subject, next)
if err == nil {
var a delivery.StepAsk
a, err = delivery.DecodeStepAsk(body)
if err == nil {
row.Oldest, row.OldestSecs = a.At, int64(now().Sub(a.At).Seconds())
}
}
if err != nil {
row.Error = fmt.Sprintf("the oldest waiting ask cannot be read: %v", err)
}
}
if row.Waiting > 0 || row.Held > 0 || row.Error != "" {
out = append(out, row)
}
}
}
return out
}
// BacklogOver reads the backlog from the bus: the worker's consumer info, and the stream's message by sequence.
func BacklogOver(js jetstream.JetStream, timeout time.Duration) *Backlog {
ctx := func() (context.Context, context.CancelFunc) {
return context.WithTimeout(context.Background(), timeout)
}
return &Backlog{
Info: func(stream, worker string) (uint64, uint64, uint64, error) {
c, cancel := ctx()
defer cancel()
s, err := js.Stream(c, stream)
if err != nil {
return 0, 0, 0, fmt.Errorf("the stream %s: %w", stream, err)
}
cons, err := s.Consumer(c, worker)
if err != nil {
return 0, 0, 0, fmt.Errorf("the worker %s: %w", worker, err)
}
info := cons.CachedInfo()
if i, err := cons.Info(c); err == nil {
info = i
}
return uint64(info.NumPending), uint64(info.NumAckPending), info.Delivered.Stream + 1, nil
},
Message: func(stream, subject string, seq uint64) ([]byte, error) {
c, cancel := ctx()
defer cancel()
s, err := js.Stream(c, stream)
if err != nil {
return nil, err
}
m, err := s.GetMsg(c, seq, jetstream.WithGetMsgSubject(subject))
if err != nil {
return nil, err
}
return m.Data, nil
},
}
}
@@ -0,0 +1,87 @@
package view
import (
"fmt"
"net/http"
"net/http/httptest"
"strings"
"testing"
"time"
"git.novox.be/novox/mesh-sdk/go/delivery"
)
func stepAsk(t *testing.T, id, kind string, at time.Time) []byte {
t.Helper()
b, err := delivery.EncodeStepAsk(delivery.StepAsk{ID: id, Kind: kind, Language: delivery.AskGo, Asker: "mesh-delivery-checker", At: at,
Check: &delivery.CheckJob{Fetch: map[string]any{"repository": "novox/x", "commit": "0123456789abcdef0123456789abcdef01234567"}, Levels: []string{delivery.LevelUnit}}})
if err != nil {
t.Fatal(err)
}
return b
}
// fakeSeats is the check worker of the go seat with three asks waiting and one held, the oldest at t0.
func fakeSeats(t *testing.T, now time.Time) *Backlog {
worker, _ := delivery.AskWorker(delivery.AskGo, delivery.AskCheck)
subject, _ := delivery.AskSubject(delivery.AskGo, delivery.AskCheck)
oldest := stepAsk(t, "m0000001aaaaaa", delivery.AskCheck, now.Add(-90*time.Second))
return &Backlog{Now: func() time.Time { return now },
Info: func(stream, w string) (uint64, uint64, uint64, error) {
switch w {
case worker:
return 3, 1, 42, nil
case "SEAT_NODE_BUILD_IMAGE_DELIVERY_worker":
return 0, 0, 1, fmt.Errorf("the worker is not there")
}
return 0, 0, 1, nil
},
Message: func(stream, s string, seq uint64) ([]byte, error) {
if s != subject || seq != 42 {
return nil, fmt.Errorf("asked %s at %d", s, seq)
}
return oldest, nil
}}
}
func TestThreeWaitingAsksAreShownWithTheOldestsAge(t *testing.T) { // 1887/A8a AC
now := time.Date(2026, 10, 23, 12, 0, 0, 0, time.UTC)
m := New()
m.SetBacklog(fakeSeats(t, now))
rows := m.Snapshot().Backlog
var check, image *BacklogRow
for i := range rows {
switch {
case rows[i].Language == "go" && rows[i].Kind == "check":
check = &rows[i]
case rows[i].Language == "image" && rows[i].Kind == "delivery":
image = &rows[i]
default:
t.Errorf("a quiet seat is listed: %+v", rows[i])
}
}
if check == nil || check.Waiting != 3 || check.Held != 1 || check.OldestSecs != 90 || !check.Oldest.Equal(now.Add(-90*time.Second)) || check.Error != "" {
t.Fatalf("check row = %+v", check)
}
if image == nil || image.Error == "" {
t.Errorf("a worker that cannot be read is dropped: %+v", rows)
}
var out strings.Builder
Table(&out, m.Snapshot())
for _, w := range []string{"BACKLOG OF BUILD ASKS", "go", "check", "3", "1m30s", "unreadable"} {
if !strings.Contains(out.String(), w) {
t.Errorf("table lacks %q:\n%s", w, out.String())
}
}
rec := httptest.NewRecorder()
Web(m).ServeHTTP(rec, httptest.NewRequest(http.MethodGet, "/", nil))
if !strings.Contains(rec.Body.String(), "BACKLOG OF BUILD ASKS") || !strings.Contains(rec.Body.String(), "1m30s") {
t.Errorf("page lacks the backlog:\n%s", rec.Body.String())
}
}
func TestNoBacklogReaderGivesNoRows(t *testing.T) {
if got := New().Snapshot().Backlog; got == nil || len(got) != 0 {
t.Errorf("backlog = %#v", got)
}
}
@@ -236,6 +236,24 @@ func differenceText(s Snapshot) [][]string {
return out
}
// backlogText is one line per language and kind: its waiting asks, the asks held by a node, the oldest waiting ask's age.
func backlogText(s Snapshot) [][]string {
var out [][]string
for _, b := range s.Backlog {
age := ""
if !b.Oldest.IsZero() {
age = (time.Duration(b.OldestSecs) * time.Second).String() + " (since " + since(b.Oldest) + ")"
}
if b.Error != "" {
age = "unreadable: " + b.Error
}
out = append(out, []string{b.Language, b.Kind, fmt.Sprint(b.Waiting), fmt.Sprint(b.Held), age})
}
return out
}
var backlogHead = []string{"LANGUAGE", "KIND", "WAITING", "HELD", "OLDEST WAITING"}
var differenceHead = []string{"NODE", "MODULE", "WANT", "SENT", "RULE", "WORDS"}
var head = []string{"NODE", "MODULE", "ASSIGNED", "HAVE", "PHASE", "HEALTH", "SINCE", "DETAIL", "WAITS"}
@@ -277,6 +295,14 @@ func Table(w io.Writer, s Snapshot) {
row(cells)
}
}
if b := backlogText(s); len(b) > 0 {
fmt.Fprintln(tw)
row([]string{"BACKLOG OF BUILD ASKS"})
row(backlogHead)
for _, cells := range b {
row(cells)
}
}
if d := differenceText(s); len(d) > 0 {
fmt.Fprintln(tw)
row([]string{"WANTS AGAINST SENDS"})
@@ -329,6 +355,7 @@ var page = template.Must(template.New("p").Parse(`<!doctype html><html><head><me
{{range .Rows}}<tr>{{range .}}<td>{{.}}</td>{{end}}</tr>{{end}}</table>
{{if .Groups}}<h2>CANDIDATES</h2><table><tr><th>GROUP</th><th>PULL REQUEST</th><th>HEAD</th><th>MODULES</th></tr>{{range .Groups}}<tr>{{range .}}<td>{{.}}</td>{{end}}</tr>{{end}}</table>{{end}}
{{if .Asks}}<h2>ASKS</h2><table><tr><th>ASK</th><th>HEADLINE</th><th>WORDS</th><th>PULL REQUESTS</th><th>ANSWER</th></tr>{{range .Asks}}<tr>{{range .Cells}}<td>{{.}}</td>{{end}}<td>{{if .Options}}<form method="post" action="/answer"><input type="hidden" name="key" value="{{.Key}}"><input name="words" placeholder="words (optional)">{{range .Options}} <button name="option" value="{{.}}">{{.}}</button>{{end}}</form>{{end}}</td></tr>{{end}}</table>{{end}}
{{if .Backlog}}<h2>BACKLOG OF BUILD ASKS</h2><table><tr>{{range .BacklogHead}}<th>{{.}}</th>{{end}}</tr>{{range .Backlog}}<tr>{{range .}}<td>{{.}}</td>{{end}}</tr>{{end}}</table>{{end}}
{{if .Differences}}<h2>WANTS AGAINST SENDS</h2><table><tr>{{range .DifferenceHead}}<th>{{.}}</th>{{end}}</tr>{{range .Differences}}<tr>{{range .}}<td>{{.}}</td>{{end}}</tr>{{end}}</table>{{end}}
{{if .Fingerprints}}<h2>FINGERPRINTS</h2><table><tr><th>MODULE</th><th>STATE</th><th>INDEXER AGAINST BUILD</th></tr>{{range .Fingerprints}}<tr>{{range .}}<td>{{.}}</td>{{end}}</tr>{{end}}</table>{{end}}
{{if .Offline}}<h2>OFFLINE</h2><table>{{range .Offline}}<tr>{{range .}}<td>{{.}}</td>{{end}}</tr>{{end}}</table>{{end}}
@@ -383,7 +410,7 @@ func Web(m *Model) http.Handler {
for _, u := range snap.Unreadable {
bad = append(bad, []string{u.Bucket, u.Key, u.Error})
}
page.Execute(w, map[string]any{"Head": head, "Rows": rows, "Groups": groupText(snap), "Asks": askPage(snap), "Fingerprints": fingerprintText(snap), "Differences": differenceText(snap), "DifferenceHead": differenceHead, "Offline": off, "Stuck": stuck, "Unreadable": bad})
page.Execute(w, map[string]any{"Head": head, "Rows": rows, "Groups": groupText(snap), "Asks": askPage(snap), "Fingerprints": fingerprintText(snap), "Differences": differenceText(snap), "Backlog": backlogText(snap), "BacklogHead": backlogHead, "DifferenceHead": differenceHead, "Offline": off, "Stuck": stuck, "Unreadable": bad})
})
}
@@ -69,6 +69,9 @@ type Snapshot struct {
// Differences is every difference between what the assigner wants a node to run and what the controller sends, with the rule that
// explains it or unexplained (novox/hq ADR 0528 clause 4, issue 1297 B8).
Differences []DifferenceRow `json:"differences"`
// Backlog is, for each language and kind of build ask with one waiting or held, the asks waiting for a node, the asks a node
// holds and the oldest waiting ask's age (issue 1887 A8a).
Backlog []BacklogRow `json:"backlog"`
}
// DifferenceRow is a module on a node whose want and send differ: the fingerprint wanted (empty when the node has no want,
@@ -186,6 +189,7 @@ type Model struct {
aasks map[string]assignerAskRecord // the assigner's asks (hold, restore), by key
chosen map[string]delivery.ViewAnswer // the answers the view wrote, by ask key
writer Writer
backlog *Backlog
answers map[string]string // by key: the router id of the asking the teller's answer is for ("" when unreadable)
newest map[string]string // module -> id
haves map[string]haveEntry
@@ -452,6 +456,25 @@ func errText(err error) string {
// Snapshot derives one row per module and node with an open assignment, a row marked left behind for each have with
// none, and a row per node's own key. Rows are ordered by node and module.
func (m *Model) Snapshot() Snapshot {
s := m.snapshot()
m.mu.Lock()
b := m.backlog
m.mu.Unlock()
s.Backlog = []BacklogRow{}
if b != nil {
s.Backlog = b.Rows() // read from the bus outside the model's lock
}
return s
}
// SetBacklog gives the model the seats' workers to read the backlog of asks from.
func (m *Model) SetBacklog(b *Backlog) {
m.mu.Lock()
defer m.mu.Unlock()
m.backlog = b
}
func (m *Model) snapshot() Snapshot {
m.mu.Lock()
defer m.mu.Unlock()
rows := []Row{}
@@ -48,6 +48,7 @@ type manifest struct {
Version string `json:"version"`
Slug string `json:"slug"`
Reads []string `json:"reads"`
Uses []string `json:"uses"`
Tools []string `json:"tools"`
OwnSecrets map[string]string `json:"own-secrets"`
SecretsOwner string `json:"secrets-owner"`
@@ -93,6 +94,10 @@ func TestManifestIsTakenAndReadsOnly(t *testing.T) { // 1134/AC7
t.Errorf("the manifest declares %q: the view writes only its answers", k)
}
}
// The view reads the build seats' streams for the backlog of asks (issue 1887 A8a): the three seats it uses.
if want := []string{"node-build-go", "node-build-image", "node-build-node"}; !reflect.DeepEqual(m.Uses, want) {
t.Errorf("uses = %v, want %v", m.Uses, want)
}
// The one state the view writes: the operator's answers to the asks that need no proof (ADR 0528 clause 6, 1297/B9), which the teller reads.
if want := []state{{Name: "answers", History: 5}}; !reflect.DeepEqual(m.State, want) {
t.Errorf("state = %+v, want %+v", m.State, want)
+5
View File
@@ -28,6 +28,11 @@
"mesh-delivery-watchman.stuck",
"node-engine.have"
],
"uses": [
"node-build-go",
"node-build-image",
"node-build-node"
],
"own-secrets": {
"broker": "${dir:state}/broker"
},
@@ -17,8 +17,6 @@ import (
"github.com/nats-io/nats.go/jetstream"
"git.novox.be/novox/mesh-sdk/go/delivery"
"mesh-delivery-watchman/internal/watch"
)
// busConf is the composed accounts file's shape (mesh-controller internal/broker): the account MESH imports the system
@@ -100,7 +98,7 @@ func TestWatchmanWhole(t *testing.T) { // 1135/AC8
{Bucket: delivery.PresenceBucket, History: 20},
{Bucket: delivery.StuckBucket, History: 20},
{Bucket: delivery.PulseBucket, LimitMarkerTTL: time.Second},
{Bucket: watch.ChecksBucket, History: 5},
{Bucket: delivery.StepsBucket("mesh-build-go"), History: 5},
} {
kv, err := js.CreateKeyValue(ctx, c)
if err != nil {
@@ -167,19 +165,26 @@ func TestWatchmanWhole(t *testing.T) { // 1135/AC8
n1.Close()
until("node n1 offline", func() bool { on, ok := online("n1"); return ok && !on })
// A check running gets a pulse, as its builder puts the key (mesh-controller CheckRunning); a builder that stops
// heartbeating lets the pulse expire, and that writes a stuck key.
running := func() []byte {
b, _ := json.Marshal(map[string]any{"head": "abc", "trunk": "def", "builder": "n1", "build": "b9", "started": time.Now().UTC()})
return b
// An ask put on a build seat gets a pulse; one no node takes lets the pulse expire, and that writes a stuck key.
askStream, _ := delivery.AskStream(delivery.AskGo)
askSubject, _ := delivery.AskSubject(delivery.AskGo, delivery.AskCheck)
putAsk := func(id string) {
t.Helper()
body, err := delivery.EncodeStepAsk(delivery.StepAsk{ID: id, Kind: delivery.AskCheck, Language: delivery.AskGo, Asker: "mesh-delivery-checker", At: time.Now().UTC(),
Check: &delivery.CheckJob{Fetch: map[string]any{"repository": "novox/x", "commit": "0123456789abcdef0123456789abcdef01234567"}, Levels: []string{delivery.LevelUnit}}})
if err != nil {
t.Fatal(err)
}
if err := ctl.Publish(askSubject, body); err != nil {
t.Fatal(err)
}
}
if _, err := kvs[watch.ChecksBucket].Put(ctx, "o.r.7", running()); err != nil {
t.Fatal(err)
}
until("a pulse for the check running", func() bool { _, ok := has(delivery.PulseBucket, pulseOf(watch.ChecksBucket, "o.r.7")); return ok })
const ask1, ask2 = "m0000001aaaaaa", "m0000002bbbbbb"
putAsk(ask1)
until("a pulse for the ask waiting", func() bool { _, ok := has(delivery.PulseBucket, pulseOf(askStream, ask1)); return ok })
var stuck delivery.Stuck
until("a stuck key for the check whose builder stopped", func() bool {
e, ok := has(delivery.StuckBucket, stuckOf(watch.ChecksBucket, "o.r.7"))
until("a stuck key for the ask no node took", func() bool {
e, ok := has(delivery.StuckBucket, stuckOf(askStream, ask1))
if !ok {
return false
}
@@ -187,7 +192,7 @@ func TestWatchmanWhole(t *testing.T) { // 1135/AC8
stuck, err = delivery.DecodeStuck(e.Value())
return err == nil
})
if stuck.Moved || stuck.Waited != "the check of o.r.7 to end" || stuck.Bucket != watch.ChecksBucket || stuck.Key != "o.r.7" {
if stuck.Moved || stuck.Bucket != askStream || stuck.Key != ask1 {
t.Fatalf("stuck = %+v", stuck)
}
// The node n1, gone offline and never back, is stuck as well.
@@ -204,17 +209,15 @@ func TestWatchmanWhole(t *testing.T) { // 1135/AC8
connect("node.n2")
until("node n2 online again", func() bool { on, ok := online("n2"); return ok && on })
until("the pulse of node n2 deleted", func() bool { _, ok := has(delivery.PulseBucket, pulseOf(delivery.PresenceBucket, "n2")); return !ok })
// A check that ends at once is the same: no stuck key, however long the bound passes.
if _, err := kvs[watch.ChecksBucket].Put(ctx, "o.r.8", running()); err != nil {
// An ask a node takes at once is the same: no stuck key, however long the bound passes.
putAsk(ask2)
until("a pulse for the second ask", func() bool { _, ok := has(delivery.PulseBucket, pulseOf(askStream, ask2)); return ok })
if _, err := kvs[delivery.StepsBucket("mesh-build-go")].Put(ctx, "n1."+ask2, []byte(`{}`)); err != nil {
t.Fatal(err)
}
until("a pulse for the second check", func() bool { _, ok := has(delivery.PulseBucket, pulseOf(watch.ChecksBucket, "o.r.8")); return ok })
if err := kvs[watch.ChecksBucket].Delete(ctx, "o.r.8"); err != nil {
t.Fatal(err)
}
until("the pulse of the second check deleted", func() bool { _, ok := has(delivery.PulseBucket, pulseOf(watch.ChecksBucket, "o.r.8")); return !ok })
until("the pulse of the second ask deleted", func() bool { _, ok := has(delivery.PulseBucket, pulseOf(askStream, ask2)); return !ok })
time.Sleep(6 * time.Second) // twice the bound and the marker time
for _, k := range []string{stuckOf(delivery.PresenceBucket, "n2"), stuckOf(watch.ChecksBucket, "o.r.8")} {
for _, k := range []string{stuckOf(delivery.PresenceBucket, "n2"), stuckOf(askStream, ask2)} {
if _, ok := has(delivery.StuckBucket, k); ok {
t.Errorf("a stuck key %s was written for a move that happened before the bound", k)
}
@@ -1,5 +1,5 @@
// Command mesh-delivery-watchman is delivery's watchman (novox/hq ADR 0429 §1, ADR 0431, ADR 0433, issue 1135): it
// hears the bus's connect and disconnect events and writes presence, mirrors every move a machine owes (a check running,
// hears the bus's connect and disconnect events and writes presence, mirrors every move a machine owes (a build ask waiting for a node,
// a node or part gone offline) into the pulse bucket with a time to live, and writes
// a stuck key when the server expires one. It holds no timer: every reaction is an event, a watch or a request at
// start or at a reconnect, and the server's expiry is the only clock.
@@ -179,6 +179,17 @@ func follow(ctx context.Context, w jetstream.KeyWatcher, each func(jetstream.Key
}
}
// stepTaken is a change of a build module's steps: a step key written for an ask is that ask taken by a node. The node's
// `paused.<node>` key and a key that is no step key are left.
func stepTaken(w *watch.Watchman, language string, e jetstream.KeyValueEntry) {
if opOf(e) != watch.Put {
return
}
if _, id, err := delivery.ParseStepKey(e.Key()); err == nil {
w.Taken(language, id)
}
}
// run is the program: it returns when ctx ends or the bus cannot be used.
func run(ctx context.Context, getenv func(string) string, logOut io.Writer) error {
logger := log.New(logOut, "mesh-delivery-watchman: ", log.LstdFlags)
@@ -223,12 +234,11 @@ func run(ctx context.Context, getenv func(string) string, logOut io.Writer) erro
}
return kv, nil
}
var kvs struct{ presence, pulse, stuck, checks jetstream.KeyValue }
var kvs struct{ presence, pulse, stuck jetstream.KeyValue }
for _, b := range []struct {
name string
into *jetstream.KeyValue
}{{delivery.PresenceBucket, &kvs.presence}, {delivery.PulseBucket, &kvs.pulse}, {delivery.StuckBucket, &kvs.stuck},
{watch.ChecksBucket, &kvs.checks}} {
}{{delivery.PresenceBucket, &kvs.presence}, {delivery.PulseBucket, &kvs.pulse}, {delivery.StuckBucket, &kvs.stuck}} {
if *b.into, err = bucket(b.name); err != nil {
return err
}
@@ -275,11 +285,22 @@ func run(ctx context.Context, getenv func(string) string, logOut io.Writer) erro
if err != nil {
return err
}
checks, err := settle(ctx, kvs.checks, func(e jetstream.KeyValueEntry) {
w.Check(e.Key(), e.Value(), e.Created(), opOf(e))
})
if err != nil {
return err
// The build seats' steps: a step key `<node>.<ask id>` written is the ask taken (issue 1887 A8b). A seat whose build module
// keeps no steps bucket (yet) is said and left; the asks of its stream are mirrored all the same.
taken := map[string]jetstream.KeyWatcher{}
for _, lang := range delivery.AskLanguages {
module, _ := delivery.BuildModuleOf(lang)
kv, err := js.KeyValue(ctx, delivery.StepsBucket(module))
if err != nil {
logger.Printf("the steps of %s cannot be read (%v): its asks are mirrored and never seen taken", module, err)
continue
}
lang := lang
sw, err := settle(ctx, kv, func(e jetstream.KeyValueEntry) { stepTaken(w, lang, e) })
if err != nil {
return err
}
taken[lang] = sw
}
// The events first, then the one request of the open connections, so no connection falls between them.
@@ -297,6 +318,23 @@ func run(ctx context.Context, getenv func(string) string, logOut io.Writer) erro
}
}
}
// The asks put on the build seats' streams: a plain subscription hears each as it is published, as the stream captures it.
for _, lang := range delivery.AskLanguages {
for _, kind := range delivery.AskKinds {
subject, _ := delivery.AskSubject(lang, kind)
lang := lang
if _, err := conn.Subscribe(subject, func(m *nats.Msg) {
a, err := delivery.DecodeStepAsk(m.Data)
if err != nil {
logger.Printf("an ask on %s is not readable: %v", m.Subject, err)
return
}
w.Asked(lang, a.ID, a.At)
}); err != nil {
return err
}
}
}
for subject, open := range map[string]bool{connectSubject: true, disconnectSubject: false} {
if _, err := conn.Subscribe(subject, onEvent(open)); err != nil {
return err
@@ -317,13 +355,14 @@ func run(ctx context.Context, getenv func(string) string, logOut io.Writer) erro
}
w.Reconcile()
errs := make(chan error, 4)
errs := make(chan error, 8)
go func() {
errs <- follow(ctx, pulses, func(e jetstream.KeyValueEntry) { w.Pulse(e.Key(), opOf(e), e.Created()) })
}()
go func() {
errs <- follow(ctx, checks, func(e jetstream.KeyValueEntry) { w.Check(e.Key(), e.Value(), e.Created(), opOf(e)) })
}()
for lang, sw := range taken {
lang, sw := lang, sw
go func() { errs <- follow(ctx, sw, func(e jetstream.KeyValueEntry) { stepTaken(w, lang, e) }) }()
}
go func() {
for {
select {
@@ -1,42 +1,42 @@
package watch
import (
"encoding/json"
"git.novox.be/novox/mesh-sdk/go/delivery"
"time"
)
// ChecksBucket is the bucket of the checks the builders are running (mesh-controller internal/broker
// CheckRunningBucket).
const ChecksBucket = "mesh-controller_checks-running"
// A waiting ask is a move (novox/hq ADR 0525 §4, issue 1887 A8b): a message on a build seat's stream (mesh-sdk go/delivery
// AskStream) that a node of the seat must pull. Its pulse is the pulse of the bucket named as the stream, under the ask's id;
// the ask put on the seat is the move beginning at its time, and the ask taken (its step key `<node>.<ask id>` written in
// the build module's steps, delivery.StepsBucket) is the move done. An ask no node takes within the bound is stuck, as any
// move. The watchman reads no bucket of the controller for it.
// checkRecord is the part of a running check's key the watchman reads, as the controller's builder writes it
// (mesh-controller internal/link/checks_nats.go CheckRunning): the head being checked.
type checkRecord struct {
Head string `json:"head"`
}
// AskedKey names a waiting ask for the pulse and stuck keys: the seat's stream and the ask's id.
func askedBucket(language string) (string, error) { return delivery.AskStream(language) }
// Check is a change in the checks-running bucket. The builder puts a check's key when the check starts and puts it
// again at every heartbeat while the check lives, and deletes it when the check ends. Each new revision of the key is
// the check moving: the pulse is deleted and written again with the full bound from that revision's time, so only a
// check whose builder stops heartbeating for the whole bound is stuck; the key deleted is the check ended. An aged-out
// key is the bucket's own expiry of a builder that died: the pulse is left to run out.
func (w *Watchman) Check(key string, value []byte, at time.Time, op Op) {
w.mu.Lock()
defer w.mu.Unlock()
switch op {
case Expired:
return
case Delete:
w.moved(ChecksBucket, key)
// Asked is an ask of a build seat of language put on its stream at time at (the ask's own time).
func (w *Watchman) Asked(language, id string, at time.Time) {
bucket, err := askedBucket(language)
if err != nil {
w.say("no pulse for the ask %s: %v", id, err)
return
}
var rec checkRecord
_ = json.Unmarshal(value, &rec)
w.mu.Lock()
defer w.mu.Unlock()
if at.IsZero() {
at = w.now()
}
if _, stalled := w.ttl(w.now().Sub(at)); !stalled {
w.stuckMoved(ChecksBucket, key)
w.mirror(bucket, id, describe("the ask %s to be taken by a node", id), "", at, 0)
}
// Taken is the ask taken by a node: its step key written. The pulse is deleted, so it never expires, and a stuck key it left is
// written as moved.
func (w *Watchman) Taken(language, id string) {
bucket, err := askedBucket(language)
if err != nil {
return
}
w.mirror(ChecksBucket, key, describe("the check of %s to end", key), rec.Head, at, 0)
w.mu.Lock()
defer w.mu.Unlock()
w.moved(bucket, id)
}
@@ -74,7 +74,7 @@ func checkValue(started time.Time, head string) []byte {
}
const (
checksKey = "mesh-controller_checks-running"
checksKey = "SEAT_NODE_BUILD_GO"
presenceNS = delivery.PresenceBucket
)
@@ -207,13 +207,13 @@ func TestRebuildTakesThePresenceOfWhoIsGone(t *testing.T) { // 1135/AC2
func TestRebuildPulsesFromCurrentKeys(t *testing.T) { // 1135/AC2
w, o, c := setup()
c.t = t0.Add(30 * time.Second)
w.Check("o.r.7", checkValue(t0, "abc"), t0.Add(10*time.Second), Put)
w.Asked("go", "askid1", t0.Add(10*time.Second))
w.SeedPresence("ace", false, t0)
w.Reconcile()
if err := w.Rebuild(func([]byte) ([]byte, error) { return connzAnswer(), nil }); err != nil {
t.Fatal(err)
}
for _, k := range []string{pulseKeyOf(t, checksKey, "o.r.7"), pulseKeyOf(t, presenceNS, "ace")} {
for _, k := range []string{pulseKeyOf(t, checksKey, "askid1"), pulseKeyOf(t, presenceNS, "ace")} {
if _, ok := o.pulses[k]; !ok {
t.Errorf("no pulse %s; have %v", k, o.pulses)
}
@@ -222,11 +222,11 @@ func TestRebuildPulsesFromCurrentKeys(t *testing.T) { // 1135/AC2
func TestReconcileDropsPulsesNothingMirrorsAnyMore(t *testing.T) { // 1135/AC2
w, o, _ := setup()
stale := pulseKeyOf(t, checksKey, "o.r.gone")
live := pulseKeyOf(t, checksKey, "o.r.7")
stale := pulseKeyOf(t, checksKey, "gone")
live := pulseKeyOf(t, checksKey, "askid1")
w.SeedPulse(stale)
w.SeedPulse(live)
w.Check("o.r.7", checkValue(t0, "abc"), t0, Put)
w.Asked("go", "askid1", t0)
w.Reconcile()
if !slices.Equal(o.pulseDeletes, []string{stale}) {
t.Fatalf("deleted = %v, want only %s", o.pulseDeletes, stale)
@@ -239,12 +239,12 @@ func TestReconcileDropsPulsesNothingMirrorsAnyMore(t *testing.T) { // 1135/AC2
func TestPulseTimeToLiveIsTheBoundMinusTheAge(t *testing.T) { // 1135/AC3
w, o, c := setup()
c.t = t0.Add(30 * time.Second)
w.Check("o.r.7", checkValue(t0, "abc"), t0, Put) // the key found at start, last renewed 30s ago
k := pulseKeyOf(t, checksKey, "o.r.7")
w.Asked("go", "askid1", t0) // the key found at start, last renewed 30s ago
k := pulseKeyOf(t, checksKey, "askid1")
if o.ttls[k] != 90*time.Second {
t.Fatalf("ttl = %v, want 90s (last renewed 30s ago)", o.ttls[k])
}
if p := o.pulses[k]; p.Bucket != checksKey || p.Key != "o.r.7" {
if p := o.pulses[k]; p.Bucket != checksKey || p.Key != "askid1" {
t.Fatalf("pulse = %+v", p)
}
w.Connect("node.ace")
@@ -257,15 +257,15 @@ func TestPulseTimeToLiveIsTheBoundMinusTheAge(t *testing.T) { // 1135/AC3
func TestAMoveAtTheBoundIsStuckAtOnce(t *testing.T) { // 1135/AC3
w, o, c := setup()
c.t = t0.Add(120 * time.Second)
w.Check("o.r.7", checkValue(t0, "abc"), t0, Put)
w.Asked("go", "askid1", t0)
if len(o.pulses) != 0 {
t.Fatalf("a pulse was written for a heartbeat 120s old: %v", o.pulses)
}
s, ok := o.stuck[stuckKeyOf(t, checksKey, "o.r.7")]
s, ok := o.stuck[stuckKeyOf(t, checksKey, "askid1")]
if !ok {
t.Fatalf("no stuck key; have %v", o.stuck)
}
if s.Moved || !s.Since.Equal(t0) || s.Waited != "the check of o.r.7 to end" {
if s.Moved || !s.Since.Equal(t0) || s.Waited != "the ask askid1 to be taken by a node" {
t.Fatalf("stuck = %+v", s)
}
}
@@ -273,13 +273,13 @@ func TestAMoveAtTheBoundIsStuckAtOnce(t *testing.T) { // 1135/AC3
func TestWaitedInWords(t *testing.T) { // 1135/AC3
w, o, c := setup()
c.t = t0.Add(200 * time.Second)
w.Check("o.r.7", checkValue(t0, "abc"), t0, Put)
w.Asked("go", "askid1", t0)
w.SeedPresence("ace", false, t0)
w.SeedPresence("part.mesh-delivery-view.ace", false, t0)
w.Reconcile()
_ = w.Rebuild(func([]byte) ([]byte, error) { return connzAnswer(), nil })
want := map[string]string{
stuckKeyOf(t, checksKey, "o.r.7"): "the check of o.r.7 to end",
stuckKeyOf(t, checksKey, "askid1"): "the ask askid1 to be taken by a node",
stuckKeyOf(t, presenceNS, "ace"): "ace to come back online",
stuckKeyOf(t, presenceNS, "part.mesh-delivery-view.ace"): "mesh-delivery-view on ace to come back online",
}
@@ -292,14 +292,14 @@ func TestWaitedInWords(t *testing.T) { // 1135/AC3
func TestEveryHeartbeatRenewsThePulse(t *testing.T) { // 1135/AC3
w, o, c := setup()
k := pulseKeyOf(t, checksKey, "o.r.7")
w.Check("o.r.7", checkValue(t0, "abc"), t0, Put)
k := pulseKeyOf(t, checksKey, "askid1")
w.Asked("go", "askid1", t0)
if o.ttls[k] != 120*time.Second {
t.Fatalf("ttl = %v, want the full 120s", o.ttls[k])
}
o.ttls[k] = 0
c.t = t0.Add(100 * time.Second)
w.Check("o.r.7", checkValue(t0, "abc"), c.t, Put) // the builder re-puts the key: same check, started long ago
w.Asked("go", "askid1", c.t) // the builder re-puts the key: same check, started long ago
if o.ttls[k] != 120*time.Second {
t.Fatalf("a heartbeat left the pulse as it was: ttl = %v, want the full 120s again", o.ttls[k])
}
@@ -308,7 +308,7 @@ func TestEveryHeartbeatRenewsThePulse(t *testing.T) { // 1135/AC3
}
// The same revision heard twice changes nothing.
o.ttls[k] = 0
w.Check("o.r.7", checkValue(t0, "abc"), c.t, Put)
w.Asked("go", "askid1", c.t)
if o.ttls[k] != 0 {
t.Fatal("one revision rewrote the pulse twice")
}
@@ -317,10 +317,10 @@ func TestEveryHeartbeatRenewsThePulse(t *testing.T) { // 1135/AC3
func TestMirroredMovesDeleteThePulse(t *testing.T) { // 1135/AC4
w, o, c := setup()
c.t = t0.Add(10 * time.Second)
w.Check("o.r.7", checkValue(t0, "abc"), t0, Put)
w.Asked("go", "askid1", t0)
w.Connect("node.ace")
w.Disconnect("node.ace")
w.Check("o.r.7", nil, c.t, Delete) // the check ended
w.Taken("go", "askid1") // the check ended
w.Connect("node.ace")
if len(o.pulses) != 0 {
t.Fatalf("pulses left: %v", o.pulses)
@@ -335,12 +335,12 @@ func TestMirroredMovesDeleteThePulse(t *testing.T) { // 1135/AC4
func TestACheckWhoseBuilderStopsIsStuckFromItsLastHeartbeat(t *testing.T) { // 1135/AC4
w, o, c := setup()
w.Check("o.r.7", checkValue(t0, "abc"), t0, Put)
w.Asked("go", "askid1", t0)
c.t = t0.Add(50 * time.Second)
w.Check("o.r.7", checkValue(t0, "abc"), c.t, Put) // the last heartbeat
w.Asked("go", "askid1", c.t) // the last heartbeat
c.t = t0.Add(170 * time.Second)
w.Pulse(pulseKeyOf(t, checksKey, "o.r.7"), Expired, c.t)
s := o.stuck[stuckKeyOf(t, checksKey, "o.r.7")]
w.Pulse(pulseKeyOf(t, checksKey, "askid1"), Expired, c.t)
s := o.stuck[stuckKeyOf(t, checksKey, "askid1")]
if !s.Since.Equal(t0.Add(50 * time.Second)) {
t.Fatalf("stuck = %+v, want since the last heartbeat", s)
}
@@ -348,35 +348,35 @@ func TestACheckWhoseBuilderStopsIsStuckFromItsLastHeartbeat(t *testing.T) { // 1
func TestAStuckKeyIsWrittenMovedNeverDeleted(t *testing.T) { // 1135/AC4
w, o, c := setup()
w.Check("o.r.7", checkValue(t0, "abc"), t0, Put)
pk := pulseKeyOf(t, checksKey, "o.r.7")
w.Asked("go", "askid1", t0)
pk := pulseKeyOf(t, checksKey, "askid1")
c.t = t0.Add(120 * time.Second)
w.Pulse(pk, Expired, c.t)
sk := stuckKeyOf(t, checksKey, "o.r.7")
sk := stuckKeyOf(t, checksKey, "askid1")
if o.stuck[sk].Moved {
t.Fatal("stuck written as moved at expiry")
}
c.t = t0.Add(300 * time.Second)
w.Check("o.r.7", nil, c.t, Delete)
w.Taken("go", "askid1")
s := o.stuck[sk]
if !s.Moved || !s.Since.Equal(t0) || s.Waited != "the check of o.r.7 to end" {
if !s.Moved || !s.Since.Equal(t0) || s.Waited != "the ask askid1 to be taken by a node" {
t.Fatalf("stuck after the move = %+v, want the same key moved", s)
}
n := o.stuckWrites
w.Check("o.r.7", nil, c.t, Delete)
w.Taken("go", "askid1")
if o.stuckWrites != n {
t.Fatal("a stuck key already moved was written again")
}
}
func TestAHeartbeatAfterStuckMovesIt(t *testing.T) { // 1135/AC4
func TestAnAskTakenAfterStuckMovesIt(t *testing.T) { // 1135/AC4
w, o, c := setup()
w.Check("o.r.7", checkValue(t0, "abc"), t0, Put)
w.Asked("go", "askid1", t0)
c.t = t0.Add(120 * time.Second)
w.Pulse(pulseKeyOf(t, checksKey, "o.r.7"), Expired, c.t)
w.Check("o.r.7", checkValue(t0, "abc"), c.t, Put) // the builder is back
if !o.stuck[stuckKeyOf(t, checksKey, "o.r.7")].Moved {
t.Fatal("a check heartbeating again is still stuck")
w.Pulse(pulseKeyOf(t, checksKey, "askid1"), Expired, c.t)
w.Taken("go", "askid1") // a node took it at last
if !o.stuck[stuckKeyOf(t, checksKey, "askid1")].Moved {
t.Fatal("an ask taken after it was stuck is still stuck")
}
}
@@ -398,23 +398,23 @@ func TestPresenceStuckThenBackOnline(t *testing.T) { // 1135/AC4
func TestExpiryMarkerWritesStuck(t *testing.T) { // 1135/AC5
w, o, c := setup()
w.Check("o.r.7", checkValue(t0, "abc"), t0, Put)
w.Asked("go", "askid1", t0)
c.t = t0.Add(120 * time.Second)
w.Pulse(pulseKeyOf(t, checksKey, "o.r.7"), Expired, c.t)
s, ok := o.stuck[stuckKeyOf(t, checksKey, "o.r.7")]
w.Pulse(pulseKeyOf(t, checksKey, "askid1"), Expired, c.t)
s, ok := o.stuck[stuckKeyOf(t, checksKey, "askid1")]
if !ok {
t.Fatalf("no stuck key; have %v", o.stuck)
}
if s.Bucket != checksKey || s.Key != "o.r.7" || s.Waited != "the check of o.r.7 to end" || !s.Since.Equal(t0) || s.Moved {
if s.Bucket != checksKey || s.Key != "askid1" || s.Waited != "the ask askid1 to be taken by a node" || !s.Since.Equal(t0) || s.Moved {
t.Fatalf("stuck = %+v", s)
}
}
func TestOwnDeleteOfAPulseWritesNothing(t *testing.T) { // 1135/AC5
w, o, c := setup()
w.Check("o.r.7", checkValue(t0, "abc"), t0, Put)
pk := pulseKeyOf(t, checksKey, "o.r.7")
w.Check("o.r.7", nil, c.t, Delete)
w.Asked("go", "askid1", t0)
pk := pulseKeyOf(t, checksKey, "askid1")
w.Taken("go", "askid1")
w.Pulse(pk, Delete, c.t) // the delete the watchman made, heard on its own watch
w.Pulse(pk, Expired, c.t)
w.Pulse("mesh-delivery-watchman_pulse.unknown.key", Expired, c.t)
@@ -427,13 +427,13 @@ func TestTheBoundCanBeShortenedForATest(t *testing.T) { // 1135/AC8
w, o, c := setup()
w.SetBound(2 * time.Second)
c.t = t0.Add(500 * time.Millisecond)
w.Check("o.r.7", checkValue(t0, "abc"), t0, Put)
if got := o.ttls[pulseKeyOf(t, checksKey, "o.r.7")]; got != 1500*time.Millisecond {
w.Asked("go", "askid1", t0)
if got := o.ttls[pulseKeyOf(t, checksKey, "askid1")]; got != 1500*time.Millisecond {
t.Fatalf("ttl = %v, want 1.5s", got)
}
c.t = t0.Add(3 * time.Second)
w.Check("o.r.8", checkValue(t0, "abc"), t0, Put)
if _, ok := o.stuck[stuckKeyOf(t, checksKey, "o.r.8")]; !ok {
w.Asked("go", "askid2", t0)
if _, ok := o.stuck[stuckKeyOf(t, checksKey, "askid2")]; !ok {
t.Fatalf("a heartbeat older than the bound is stuck at once; have %v", o.stuck)
}
}
@@ -81,7 +81,7 @@ func TestManifestStateAndReads(t *testing.T) { // 1135/AC6
}
reads := slices.Clone(m.Reads)
slices.Sort(reads)
if !reflect.DeepEqual(reads, []string{"mesh-controller.checks-running"}) {
if !reflect.DeepEqual(reads, []string{"mesh-build-go.steps", "mesh-build-image.steps", "mesh-build-node.steps"}) {
t.Errorf("reads = %v", reads)
}
for _, k := range []string{"emits", "context", "items", "consumes", "provides", "claims"} {
+3 -1
View File
@@ -3,7 +3,9 @@
"version": "1",
"slug": "watchman",
"reads": [
"mesh-controller.checks-running"
"mesh-build-go.steps",
"mesh-build-image.steps",
"mesh-build-node.steps"
],
"state": [
{ "name": "presence", "history": 20 },