fix(flush): the gate settles its waiter from the machine, never from a chain
Some checks failed
CI / Node 22 (push) Successful in 12m26s
CI / Node 24 (push) Successful in 12m19s
CI / Bun (latest) (push) Successful in 12m28s
CI / Integration + conformance (Node 22) (push) Failing after 17m12s
Some checks failed
CI / Node 22 (push) Successful in 12m26s
CI / Node 24 (push) Successful in 12m19s
CI / Bun (latest) (push) Successful in 12m28s
CI / Integration + conformance (Node 22) (push) Failing after 17m12s
The single-flight gate queued its follow-up as `leader.catch().then(() => this.flush())`. That waiter is settled ONLY by resolving the very promise the leader is being awaited through, so the moment anything inside a flush body awaits flush(), the promise graph closes on itself and nobody resolves — an unbounded hang, not a slow flush, presenting exactly like a bulk write timing out. No current call site awaits a flush from inside one, so this is a latent cycle rather than an observed one; the gate should not depend on that staying true. The queue is now a bare deferred. The leader's finally opens the gate and PROMOTES the waiter to a new leader, settling the deferred from that run; the finally returns nothing, so the leader never awaits its own follower. Every exit runs the same promotion — the leader resolving, the leader rejecting, the promoted run rejecting — so a queued caller is settled exactly once on every path, and a synchronous failure starting the promoted run is reported to the waiter instead of thrown into the leader's finally. close() drains both handles. tests/unit/brainy/flush-single-flight.test.ts pins the invariant on each path that must settle a waiter: many callers during one flush all resolve within a bound (one body, one follow-up, peak concurrency 1); a REJECTING leader still runs and settles the queued waiter; a rejecting follow-up settles its waiter and leaves the gate open; and the leader returns without waiting for a deliberately slower follower.
This commit is contained in:
parent
ebb3a4bf13
commit
dea3ec2031
3 changed files with 243 additions and 18 deletions
|
|
@ -785,9 +785,25 @@ export class Brainy<T = any> implements BrainyInterface<T> {
|
|||
* "Flushing Brainy indexes and caches to disk..." runs overlapping 3s
|
||||
* apart on one brain, their walls growing 295ms → 4.9s as they contended
|
||||
* for the same providers.
|
||||
*
|
||||
* THE WAITER IS SETTLED BY THE MACHINE, NEVER BY A PROMISE CHAIN. The queue
|
||||
* is a BARE DEFERRED (`_flushQueued` plus its `_flushQueuedSettle` handles),
|
||||
* not `leader.then(() => this.flush())`. A chained follow-up is settled only
|
||||
* by resolving the very promise the leader is being awaited through, so the
|
||||
* moment anything inside a flush body awaits `flush()` the graph closes on
|
||||
* itself and NOBODY resolves — an unbounded hang, not a slow flush. Here the
|
||||
* leader never awaits the queue: its `finally` PROMOTES the waiter to a new
|
||||
* leader and settles the deferred from that run, and the leader's own
|
||||
* promise settles without waiting for it. Every exit — the leader
|
||||
* resolving, the leader REJECTING, the promoted run rejecting — runs the
|
||||
* same promotion, so a queued caller is always settled exactly once.
|
||||
*/
|
||||
private _flushInFlight: Promise<void> | null = null
|
||||
private _flushFollowUp: Promise<void> | null = null
|
||||
private _flushQueued: Promise<void> | null = null
|
||||
private _flushQueuedSettle: {
|
||||
resolve: () => void
|
||||
reject: (error: unknown) => void
|
||||
} | null = null
|
||||
/** Flush bodies that got past the single-flight gate (pinned by tests). */
|
||||
private _flushBodyRuns = 0
|
||||
/** Flush bodies running right now, and the high-water mark — which the
|
||||
|
|
@ -12987,29 +13003,61 @@ export class Brainy<T = any> implements BrainyInterface<T> {
|
|||
// crossed BEFORE any await, so two callers in the same tick cannot both
|
||||
// find the field empty.
|
||||
if (this._flushInFlight) {
|
||||
if (!this._flushFollowUp) {
|
||||
// The running flush's failure is not this follow-up's failure: it is
|
||||
// reported to ITS caller, and the queued work still gets its turn.
|
||||
this._flushFollowUp = this._flushInFlight
|
||||
.catch(() => {})
|
||||
.then(() => {
|
||||
this._flushFollowUp = null
|
||||
return this.flush()
|
||||
})
|
||||
if (!this._flushQueued) {
|
||||
// A BARE DEFERRED, not a chain off the leader — see the field's doc.
|
||||
// Nothing here awaits the leader, so no waiter can ever be reachable
|
||||
// only through the promise it is itself blocking.
|
||||
this._flushQueued = new Promise<void>((resolve, reject) => {
|
||||
this._flushQueuedSettle = { resolve, reject }
|
||||
})
|
||||
}
|
||||
return this._flushFollowUp
|
||||
return this._flushQueued
|
||||
}
|
||||
return this.startFlushLeader()
|
||||
}
|
||||
|
||||
/**
|
||||
* @description Run one flush body as the leader and install it as
|
||||
* `_flushInFlight`. On settle — resolved OR rejected — the gate opens and
|
||||
* the ONE queued waiter (if any) is promoted. The `finally` callback returns
|
||||
* nothing on purpose: a callback that returned the promoted run's promise
|
||||
* would make the leader await its own follower.
|
||||
* @returns The leader's own promise, settling on its own body alone.
|
||||
*/
|
||||
private startFlushLeader(): Promise<void> {
|
||||
const run = this._runFlush()
|
||||
// `finally` and not `then`: a failed flush must still open the gate, or
|
||||
// one rejection would wedge every later flush behind a promise nobody
|
||||
// will ever settle.
|
||||
const gated = run.finally(() => {
|
||||
const gated: Promise<void> = run.finally(() => {
|
||||
if (this._flushInFlight === gated) this._flushInFlight = null
|
||||
this.promoteQueuedFlush()
|
||||
})
|
||||
this._flushInFlight = gated
|
||||
return gated
|
||||
}
|
||||
|
||||
/**
|
||||
* @description Promote the single queued waiter (if one is waiting) to
|
||||
* leader and settle its deferred from that run. Never throws into the
|
||||
* leader's `finally`: a synchronous failure starting the promoted run is
|
||||
* reported to the waiter, which must be settled on every path.
|
||||
* @returns Nothing.
|
||||
*/
|
||||
private promoteQueuedFlush(): void {
|
||||
const settle = this._flushQueuedSettle
|
||||
if (!settle) return
|
||||
// Clear BEFORE starting, so the promoted run's own joiners queue afresh
|
||||
// rather than joining a deferred that is already being settled.
|
||||
this._flushQueued = null
|
||||
this._flushQueuedSettle = null
|
||||
try {
|
||||
this.startFlushLeader().then(settle.resolve, settle.reject)
|
||||
} catch (error) {
|
||||
settle.reject(error)
|
||||
}
|
||||
}
|
||||
|
||||
/**
|
||||
* @description The flush body — everything {@link flush} promises, run
|
||||
* exactly once at a time by that method's single-flight gate. Private
|
||||
|
|
@ -20409,9 +20457,11 @@ export class Brainy<T = any> implements BrainyInterface<T> {
|
|||
// awaits its leader too, so the second pass is a no-op unless a writer
|
||||
// raced this close.
|
||||
for (let pass = 0; pass < 2; pass++) {
|
||||
const chain = this._flushFollowUp ?? this._flushInFlight
|
||||
if (!chain) break
|
||||
await chain.catch(() => {})
|
||||
const inFlight = this._flushInFlight
|
||||
const queued = this._flushQueued
|
||||
if (!inFlight && !queued) break
|
||||
if (inFlight) await inFlight.catch(() => {})
|
||||
if (queued) await queued.catch(() => {})
|
||||
}
|
||||
|
||||
// Cancel any pending post-import background deduplication FIRST — it is a
|
||||
|
|
|
|||
Reference in a new issue