Retire the controller's merges consumer and read merges from the forge bridge's stream, so a held merge can never again queue every merge or look like a deaf controller (issue 774)

The controller-merges consumer (no max-deliver, one at a time) held the de-spiegel merge e25e1beb from
10-29 09:39: keepForgeMerge put an announcement with its Paths into delivery-merges (16 KiB values), got
10054, retried "not now" forever, and 167 merges queued behind it. Every five-minute hold outran the deaf
bound, so core.controller.deaf flapped.

Merges are mesh-delivery's: the gate merges on the Approve and the forge bridge says each merge on its
stream FORGE (pull request, head, merge commit), kept a week. The controller's copy was a leftover, read
only at a moved module's registration (ADR 0420 §7, ADR 0432). That read now finds the merge on FORGE;
the consumer, its follow, its handler and its hold delays are removed, and every assertion removes the
retired consumer from the bus.

Kind fix for the watchdog: the event loop records when a message it held back is due again, and deafness
counts from then, so no hold longer than the bound (on any consumer) reads as deafness.

Quick fix: strip Paths from the kept announcement and add max-deliver with a dead letter. Thorough fix:
remove the duplicate duty and make the deaf watchdog hold-aware. Chose thorough: the controller is being
deprecated and the merges already have an owner; the hold-vs-deaf confusion is a kind, not one consumer.

Research 072 piece: 03-domains (Control shrinks to placement and composition; the delivery parts own
what a merge is) and milestone 3 (health signals stop sharing the controller's loop assumptions).
This commit is contained in:
2026-10-30 18:05:46 +01:00
parent a4426190b0
commit 48d9f22fa5
19 changed files with 348 additions and 385 deletions
+6 -20
View File
@@ -62,7 +62,7 @@ func Nats(js *broker.JetStream) Inbound {
// whatever was asked for — and not at all when nothing was.
func (n *natsInbound) Also(kind string) error {
switch kind {
case KindModuleMoved, KindCatchUp, KindSourceMoved, KindProvisioner, KindPullUpdated, KindDecided,
case KindModuleMoved, KindCatchUp, KindProvisioner, KindPullUpdated, KindDecided,
KindDeliveryMerged:
n.follows[kind] = true
return nil
@@ -145,24 +145,6 @@ func (n *natsInbound) Receive(ctx context.Context, act func(context.Context, Con
defer func() { _ = followed.Unsubscribe() }()
}
// And the forge's merges, off a consumer of their own: no max-deliver and no reset, so a merge held for the
// plans waits for them however long (novox/hq ADR 0407 decision 3). Bound to the same channel: one loop.
if n.follows[KindSourceMoved] {
if events == nil {
events = make(chan *nats.Msg, Prefetch)
}
merges, err := standingBy(ctx, log.Default(), "EVENTS", func() (*nats.Subscription, error) {
return js.ChanSubscribe(MergedSubject, events, nats.Bind("EVENTS", broker.MergesConsumerName))
})
if err != nil {
return fmt.Errorf("subscribing to the forge's merges: %w", err)
}
if merges == nil {
return nil
}
defer func() { _ = merges.Unsubscribe() }()
}
// A connection that dropped is said, not discovered. A controller whose bus connection is gone
// is a mesh where nothing can be told anything.
gone := make(chan error, 1)
@@ -423,7 +405,11 @@ func (m *natsControl) Hold(after time.Duration) error {
if _, already := m.on.since[m.seq]; !already {
m.on.since[m.seq] = time.Now()
}
return m.msg.NakWithDelay(after)
if err := m.msg.NakWithDelay(after); err != nil {
return err
}
Loop.HeldUntil(time.Now().Add(after))
return nil
}
func (m *natsControl) forget() {