- Add SessionCache (LRU + TTL, per-API-key scoped) mapping
conversation-prefix hash -> upstream Lingma sessionId.
- Hash only user/system/developer turns so client-side
assistant reformatting doesn't invalidate the key.
- On cache hit: reuse sessionId, send only the latest user
message with isReply=true, and stick the request to the
instance that originally served it.
- LingmaGatewayClient.chat_complete/chat_stream accept
session_id/is_reply and report the real finish.sessionId
via out_meta so we persist what Lingma actually allocated.
- Invalidate cache on non-stream failure; skip writes on
cancelled/partial streams.
- Expose cache stats in /internal/stats and /metrics.
- Configurable via SESSION_REUSE_ENABLED / SESSION_CACHE_MAX_ENTRIES
/ SESSION_CACHE_TTL_SEC (documented in README + .env.example).
Made-with: Cursor
Lingma streams answers via chat/answer + chat/finish notifications and
never sends a JSON-RPC response for chat/ask. The old code awaited
rpc.request("chat/ask") and swallowed the TimeoutError, so every chat
was forced to wait the full rpc_timeout (default 30s) before draining
the stream queue - even though the first token was already present in
the queue within ~2s.
Effect:
- non-stream TTFB dropped from ~30s to actual upstream latency (~2-3s).
- stream first-chunk dropped from ~30s to upstream first-token latency.
- consume_stream idle timeout decoupled from rpc_timeout so shortening
rpc_timeout no longer starves long completions.
Switch chat/ask to rpc.notify (fire-and-forget) and rely entirely on the
existing chat/answer + chat/finish handlers for result delivery.
Made-with: Cursor
Behavior hardening (M1):
- Fix `_chat_streams` memory leak: pop_stream on completion, error, and
client disconnect.
- Add WebSocket reconnect with state machine (stopped/starting/ready/
reconnecting/failed/closed) and exponential backoff, so a Lingma
restart no longer requires restarting the gateway.
- Lazy initialization: startup failure is non-fatal, first real request
triggers retry, `/healthz` reflects readiness.
- Migrate FastAPI on_event to lifespan.
- Structured JSON logging with request_id ContextVar; `x-request-id`
propagated to responses.
- SSE now sets `Cache-Control: no-cache`, `X-Accel-Buffering: no` to
defeat proxy buffering.
- OpenAI schema compatibility: `content` accepts str | list[parts] | None,
added `developer`/`function` roles, `tools/tool_choice/stream_options/
user/max_tokens` fields, and `stream_options.include_usage` emits final
usage chunk.
- `require_bearer` uses `hmac.compare_digest`; `/metrics` now requires
Bearer when `METRICS_TOKEN` or `API_KEYS` are set.
- Python 3.10/3.11 `TimeoutError` vs `asyncio.TimeoutError` unified.
- Error responses no longer leak `auto_login.status()` details.
Backpressure (M2 / A2):
- New `InFlightGuard` with per-request ticket, queue + rejection
accounting, `BackpressureRejected` raises 429 + `Retry-After` once
`GATEWAY_QUEUE_TIMEOUT_SEC` elapses.
- Streaming ticket ownership transfers to the generator so CancelledError
from client disconnect still releases the slot.
- `/internal/stats.concurrency` and `/metrics` expose in_flight/queued/
accepted_total/rejected_total/max_in_flight.
Multi-instance pool (M2 / A1 + B3):
- New `LingmaPool` with N processes, each with its own workDir, socket
port (dynamic when N>1), and `AutoLoginManager`.
- Account parser supports CSV (`u1:p1,u2:p2`) and JSON formats via
`LINGMA_ACCOUNTS`; falls back to `LINGMA_USERNAME/LINGMA_PASSWORD` for
backwards compatibility (N=1 keeps legacy paths/ports).
- Routing: sticky affinity by `user` / system-prompt hash, then
least-in-flight, finally round-robin fallback for unhealthy pool.
- `/healthz` reports per-instance state and ready count.
- `/internal/stats.pool` and `/metrics` expose per-instance
`gateway_pool_instance_in_flight{name}` / `gateway_pool_instance_ready{name}`.
- `/internal/auto-login/start?instance=inst-N` targets a specific instance;
`/internal/auto-login/status` lists all instances.
Compat notes:
- `.env.example` adds `METRICS_TOKEN`, `LOG_LEVEL`, `GATEWAY_MAX_IN_FLIGHT`,
`GATEWAY_QUEUE_TIMEOUT_SEC`, `LINGMA_ACCOUNTS`, `LINGMA_INSTANCE_COUNT`.
- `.gitignore` cleaned up data/ duplication.
- Existing single-instance deployments keep working without config change.
Made-with: Cursor
