Codex/OpenAI upstream switched from a WebSocket pool to HTTP/2. The
outbound H2 transport set neither ReadIdleTimeout nor ResponseHeaderTimeout
(the OpenAI profile forces ResponseHeaderTimeout=0), so a pooled H2
connection silently killed by a proxy/NAT becomes a "dead connection":
both ends believe it is alive and a request assigned to it hangs until the
OS TCP retransmit timeout (minutes) before the first byte — observed as an
8m37s TTFT with an eventual 200. Occasional (only when a request lands on a
dead pooled conn) and across all groups (shared OpenAI transport); worse on
larger idle pools.
Explicitly configure http2 on the openai_h2 transport and enable active
PING health checks (ReadIdleTimeout=15s, PingTimeout=15s) so dead
connections are detected and evicted at the source, instead of relying on
ResponseHeaderTimeout as an after-the-fact backstop. Scoped to the
openai_h2 path only; Claude/Gemini (default) and h1 modes are untouched.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
OpenAI OAuth (ChatGPT) accounts deterministically reject plan-gated
models with 400 "The 'X' model is not supported when using Codex with
a ChatGPT account". Account selection had no capability filtering for
this, so the scheduler kept picking the same account for the same model
forever; every attempt burned an upstream call and surfaced to clients
as a retryable 502, sustaining client retry storms.
Treat this 400 like upstream model-not-found: mark the (account, model)
pair via SetModelRateLimit (30min cooldown) so
IsSchedulableForModelWithContext skips the account for that model
during selection, and return true so the in-flight request fails over
to another account through the existing UpstreamFailoverError path.
Expose the active weekly subscription window through /v1/usage, calculate offsets with the same normalized page size used by queries, and keep user-facing date ranges on the browser's local calendar date.
Constraint: Preserve existing response fields and avoid new dependencies
Rejected: Keep duplicate inline date formatters | a shared local-date utility prevents the same UTC regression in both views
Confidence: high
Scope-risk: narrow
Reversibility: clean
Directive: Keep Offset and Limit based on the same normalized page size
Tested: go test ./internal/pkg/pagination ./internal/handler; go vet ./internal/pkg/pagination ./internal/handler; frontend 923 tests; pnpm typecheck; pnpm lint:check; pnpm build
Not-tested: Live API request against a deployed subscription
Related: #4121
Direct Go/Docker deployments were serving Vite-hashed /assets without
Cache-Control, so browsers re-downloaded the console on every visit.
Align with deploy/Caddyfile while keeping index.html on no-cache.
Closes#4129
Allow Grok API-key accounts in Responses forwarding and connection tests, expose creation and edit defaults in the dashboard, and document the supported setup.
