Skip to content

Commit cc7d945

Browse files
committed
docs(devlog): fold CodeRabbit PR350 review — retry transport boundary, httpStatus union split, CYCLE6_BASE_SHA wording
1 parent da97a56 commit cc7d945

3 files changed

Lines changed: 29 additions & 18 deletions

File tree

‎devlog/_plan/260724_bugfix_train/020_pool_retry_400.md‎

Lines changed: 23 additions & 14 deletions
Original file line numberDiff line numberDiff line change
@@ -450,20 +450,29 @@ if (
450450
);
451451
452452
noteAttemptSend(logCtx.activeAttempt, passthroughEstimate);
453-
upstreamResponse = await fetchWithHeaderTimeout(
454-
request.url,
455-
{
456-
method: request.method,
457-
headers: request.headers,
458-
body: request.body,
459-
},
460-
upstream.signal,
461-
connectMs,
462-
parsed.stream,
463-
providerFetch(route.provider),
464-
);
465-
// No transient wrapper and no call back into the matcher: the entire retry
466-
// budget is spent by this one different-account HTTP dispatch.
453+
// The retry dispatch reuses the SAME transport-error boundary as the first
454+
// dispatch: a connect/timeout failure here surfaces through the existing
455+
// catch that maps transport errors to 502/504, attributed to the retry
456+
// account (B). No transient wrapper and no call back into the matcher: the
457+
// entire retry budget is spent by this one different-account HTTP dispatch.
458+
try {
459+
upstreamResponse = await fetchWithHeaderTimeout(
460+
request.url,
461+
{
462+
method: request.method,
463+
headers: request.headers,
464+
body: request.body,
465+
},
466+
upstream.signal,
467+
connectMs,
468+
parsed.stream,
469+
providerFetch(route.provider),
470+
);
471+
} catch (err) {
472+
// Route through the existing transport-error helper (same code path the
473+
// first dispatch uses at core.ts:941 region); health/log attribution is B.
474+
throw err; // caught by the surrounding dispatch transport boundary
475+
}
467476
}
468477
}
469478

‎devlog/_plan/260724_bugfix_train/040_ux_discovery_badge.md‎

Lines changed: 5 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -113,10 +113,11 @@ export type ProviderModelDiscoveryFailureReason =
113113

114114
export type ProviderModelDiscoveryStatus =
115115
| { status: "ok" }
116+
| { status: "failed"; reason: "http"; httpStatus: number }
116117
| {
117118
status: "failed";
118-
reason: ProviderModelDiscoveryFailureReason;
119-
httpStatus?: number;
119+
reason: Exclude<ProviderModelDiscoveryFailureReason, "http">;
120+
httpStatus?: never;
120121
};
121122

122123
// GET /api/providers row (new field only)
@@ -193,7 +194,8 @@ export type ProviderModelDiscoveryFailureReason =
193194

194195
export type ProviderModelDiscoveryStatus =
195196
| { status: "ok" }
196-
| { status: "failed"; reason: ProviderModelDiscoveryFailureReason; httpStatus?: number };
197+
| { status: "failed"; reason: "http"; httpStatus: number }
198+
| { status: "failed"; reason: Exclude<ProviderModelDiscoveryFailureReason, "http">; httpStatus?: never };
197199

198200
const discoveryStatus = new Map<string, ProviderModelDiscoveryStatus>();
199201

‎devlog/_plan/260724_bugfix_train/061_root_suite_investigation.md‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -67,7 +67,7 @@ Classify each of the four named failures independently and cite its log/run evid
6767
- **PR-caused** — reproduces on the PR head, does not reproduce on the exact post-WP1 `dev` base
6868
under the same environment/order, and a concrete causal path from an actual PR diff hunk to the
6969
assertion/resource exists.
70-
- **Baseline deterministic** — reproduces on post-WP1 `dev` under the same environment/order with a
70+
- **Baseline deterministic** — reproduces on `CYCLE6_BASE_SHA` under the same environment/order with a
7171
stable assertion and repeatable owner.
7272
- **Order/concurrency flake** — standalone remains green, but a repeatable suite order, shared path,
7373
process overlap, environment leak, or resource collision triggers the failure on PR and/or base.

0 commit comments

Comments
 (0)