fix(dynamic-truncator): bound session.messages fetch to stop forever-hang on Read (#4086)

Root cause: `getContextWindowUsage` caches the *promise* of
`fetchContextWindowUsage` in a per-session WeakMap keyed by client. When
`ctx.client.session.messages({ path: { id: sessionID } })` never settles
(observed once `service=session.processor ... error=Aborted process`
takes hold), the cached pending promise wedges every concurrent and
later caller in the same session. The five hooks that share one
`createDynamicTruncator(ctx)` -- directory-agents-injector,
directory-readme-injector, rules-injector, tool-output-truncator, plus
indirect callers -- all await that same poisoned promise on every Read,
so the user-facing tool chain hangs forever and ESC cannot break it.
Reporters in #4086 land on this path consistently when reading AGENTS.md
files (which trigger directory-agents-injector via the directory walk).

Fix: race the underlying `session.messages` call against a 5s timeout
through a new `withFetchTimeout` helper. On timeout the catch block logs
the failure and returns `null`, which `dynamicTruncate` already treats
as the "context usage unavailable" signal and falls back to the static
truncation budget. Successful responses still cache as before. The
`message.updated finish=true` invalidation hook still clears poisoned
caches on the next completed turn so retries are clean.

Tests:
- Add a never-settling `session.messages` mock with a 50 ms override via
  the new `_setContextWindowUsageFetchTimeoutMsForTesting` hook (matches
  the established `_setXxxForTesting` pattern in `opencode-http-api.ts`
  and `prompt-async-gate.ts`).
- Three new BDD cases pin the fix: (1) single caller returns null fast,
  (2) parallel concurrent callers all unblock on the same cached promise
  instead of hanging, (3) invalidate + retry rehydrates cleanly.
- All 8 pre-existing tests in the file still pass (happy paths, cache
  reuse, invalidation, env/model fallback).

Verification:
- `bun test src/shared/dynamic-truncator.test.ts` -- 11 pass.
- `bun test src/shared/prompt-async-route-audit.test.ts` -- 6 pass
  (added log import, no raw prompt route added).
- `bun test` (full suite) -- 7009 pass, 1 skip, 1 pre-existing flake in
  `closeTmuxPane` mock.module test (reproduces on dev without this
  change; isolated run passes).
- `bun run typecheck` -- clean.
- `bun run build` -- clean (esm bundle + tsc + schema).
- Manual harness `.debugging/manual-qa.ts` (uncommitted) drives the same
  shape as the real hook chain and resolves the hang scenario in 51 ms.
This commit is contained in:
YeonGyu-Kim
2026-05-17 03:51:56 +09:00
parent 169e61f775
commit 67ead7bf6d
2 changed files with 167 additions and 5 deletions
+113 -1
View File
@@ -2,7 +2,11 @@
import { describe, expect, it, afterEach } from "bun:test"
import { getContextWindowUsage, invalidateContextWindowUsageCache } from "./dynamic-truncator"
import {
_setContextWindowUsageFetchTimeoutMsForTesting,
getContextWindowUsage,
invalidateContextWindowUsageCache,
} from "./dynamic-truncator"
const ANTHROPIC_CONTEXT_ENV_KEY = "ANTHROPIC_1M_CONTEXT"
const VERTEX_CONTEXT_ENV_KEY = "VERTEX_ANTHROPIC_1M_CONTEXT"
@@ -89,6 +93,114 @@ function createCountingContextUsageMockContext(inputTokens: number) {
describe("getContextWindowUsage", () => {
afterEach(() => {
resetContextLimitEnv()
_setContextWindowUsageFetchTimeoutMsForTesting(undefined)
})
describe("#given client.session.messages never settles", () => {
describe("#when getContextWindowUsage is called with a fast fetch timeout", () => {
it("#then returns null instead of hanging forever", async () => {
// given
_setContextWindowUsageFetchTimeoutMsForTesting(50)
const ctx = {
client: {
session: {
messages: () => new Promise<never>(() => {}),
},
},
}
// when
const start = Date.now()
const usage = await getContextWindowUsage(ctx as never, "ses_hang_messages", {
anthropicContext1MEnabled: false,
})
const elapsed = Date.now() - start
// then
expect(usage).toBeNull()
expect(elapsed).toBeLessThan(2000)
})
it("#then a parallel concurrent caller also resolves to null instead of hanging on the cached promise", async () => {
// given
_setContextWindowUsageFetchTimeoutMsForTesting(50)
const ctx = {
client: {
session: {
messages: () => new Promise<never>(() => {}),
},
},
}
// when
const start = Date.now()
const [first, second] = await Promise.all([
getContextWindowUsage(ctx as never, "ses_hang_messages_parallel", {
anthropicContext1MEnabled: false,
}),
getContextWindowUsage(ctx as never, "ses_hang_messages_parallel", {
anthropicContext1MEnabled: false,
}),
])
const elapsed = Date.now() - start
// then
expect(first).toBeNull()
expect(second).toBeNull()
expect(elapsed).toBeLessThan(2000)
})
it("#then a follow-up call after invalidation retries fresh instead of being poisoned by the timeout", async () => {
// given
_setContextWindowUsageFetchTimeoutMsForTesting(50)
let messagesCalls = 0
let shouldHang = true
const ctx = {
client: {
session: {
messages: () => {
messagesCalls += 1
if (shouldHang) {
return new Promise<never>(() => {})
}
return Promise.resolve({
data: [
{
info: {
role: "assistant",
providerID: "anthropic",
modelID: "claude-sonnet-4-5",
tokens: {
input: 100000,
output: 0,
reasoning: 0,
cache: { read: 0, write: 0 },
},
},
},
],
})
},
},
},
}
// when
const firstUsage = await getContextWindowUsage(ctx as never, "ses_hang_then_recover", {
anthropicContext1MEnabled: false,
})
invalidateContextWindowUsageCache(ctx as never, "ses_hang_then_recover")
shouldHang = false
const secondUsage = await getContextWindowUsage(ctx as never, "ses_hang_then_recover", {
anthropicContext1MEnabled: false,
})
// then
expect(firstUsage).toBeNull()
expect(secondUsage?.remainingTokens).toBe(100000)
expect(messagesCalls).toBe(2)
})
})
})
it("uses 1M limit when model cache flag is enabled", async () => {