Repository navigation
Conversation
… engine, TLS on a bundled BoringSSL, lent bodies The entry's HTTP/2 listeners ran on Node's http2 module, the only part of it not served by the native engine, and its TLS listener on the OpenSSL inside node. @morojs/engine 1.1.19 speaks HTTP/2 itself (RFC 9113 and HPACK, its own implementation): ALPN h2 on the TLS listener, prior knowledge on the cleartext one, HTTP/1.1 beside it on the same port. A request on a stream reaches the same routes, middleware and response objects as an HTTP/1.1 request, so the two listeners lose their `engine: 'node'` and nothing else changes; they now cluster on worker threads like the rest. /baseline2 no longer sets its content type by hand, which was only there for the Node HTTP/2 server: the framework's prepared reply answers it on a stream as it does on a connection. @morojs/engine 1.1.20 terminates TLS with a BoringSSL it carries inside its binary (kept current by a scheduled engine release for every upstream snapshot; node's OpenSSL remains one setting away and is chosen by itself under FIPS or an OpenSSL config). Nothing in the entry selects it: it is the engine's default. `server.lendBodies: true` (MoroJS 1.8.23): a body the framework does not parse is lent by the engine for the length of the request instead of handed over, so the 10 KiB bodies of 8gbit no longer count against V8's external memory until a collection frees them. JSON, form and text bodies are read as strings in the first place, and the registry of in-flight responses no longer drags finished requests into the old generation (the async profile). Measured locally at one worker on one core, CPU per request: json-tls 16.7 to 15.1 us and 8gbit 17.6 to 15.4 us from BoringSSL alone (21.5 before lent bodies), the delayed reply at 64 connections 25.1 to 22.9, the JSON route behind its compression middleware 15.1 to 12.5. h2load at 64 connections of 100 streams against the previous image: baseline-h2c 46,400 to 1,150,000 requests a second, baseline-h2 39,700 to 1,050,000, json-h2c 38,000 to 167,000, static-h2 34,800 to 63,000.
|
/benchmark -f morojs --save |
|
👋 Benchmark request received. A collaborator will review and approve the run. |
Benchmark ResultsFramework:
Composite score, main against this run, in the board's default view:
The results on main were measured in another session, so a few points up or down can be the machine and not the change. Full log |
|
/benchmark -f morojs -t production-stack --save |
|
👋 Benchmark request received. A collaborator will review and approve the run. |
Benchmark ResultsFramework: No results captured Full log |
|
@MDA2AV The production-stack re-run here failed before measuring anything: the shared authsvc container exited (0) with no output, same as roadrunner-gateway on #1757 and tako-gateway on #1762 today, so it looks like the bench host. The full run from yesterday saved fine and those results are in the PR. Happy for this to merge as is; I'll re-run production-stack once the sidecar is sorted. |
…uthsvc on Rust 1.100, unbreak the aspnet gateway entries (#1767) * validate.sh: stop the standalone container before the compose profiles The standalone container runs on --network host whenever an entry subscribes to a compose profile, and it was left running while the gateway and production-stack stacks came up. An entry that also serves h2/h3 keeps 8443 (and 8080), so the stack's edge failed to bind, shared the port, or exited on the QUIC port, and the checks reached the standalone server instead of the stack. * authsvc: ship the real binary under cargo 1.100's build-dir layout production-stack has failed on the bench host since the morning of 2026-11-13 on every entry (roadrunner-gateway #1757, morojs #1761, tako-gateway #1762): the shared authsvc container exits 0 with no output, the edge never sees it healthy, and the run ends with no results. rust:1-alpine moved to Rust 1.100 on 2026-11-12. Its cargo keeps each package's artifacts under target/release/build/<package>/<hash>/ rather than target/release/deps/, so the cache-priming step's `rm -rf target/release/deps/authsvc*` stopped matching anything. The placeholder built from `fn main() {}` stayed up to date, the second `cargo build` finished in 0.04s without compiling src/, and the image shipped the placeholder: a 418 KB binary that exits 0 at once. `cargo clean --release -p authsvc` removes the placeholder's artifacts wherever cargo keeps them, and a grep for the server's startup line fails the build if the placeholder is ever shipped again. Built against the current rust:1-alpine (1.100.0): a 1.2 MB binary that logs "authsvc listening on ...", answers /_health with 200 and an /_auth without a token with 401. * aspnet-minimal_nginx, aspnet-minimal_caddy: build against aspnet-minimal's Services and Types Both Dockerfiles still COPY aspnet-minimal's AppData.cs and Models.cs, which #1002 (2026-07-18) split into Services/ and Types/, so neither image has built since. They were the only failures in #1615's validation sweep, and their published results have not been refreshed since July. Copy Handlers.cs with Services/ and Types/, register DatabaseService, DatasetService, ItemService and the AppJsonContext resolver the way aspnet-minimal's Program.cs does, and give the nginx entry's /api routes DatabaseService's Postgres pool in place of AppData.PgDataSource. The caddy entry gains the StackExchange.Redis reference DatabaseService needs; it connects to nothing unless REDIS_URL is set, which gateway-h3 never does. --------- Co-authored-by: Jacob Rothstein <[email protected]>
|
Two of the failures here came from the harness, not morojs, and #1767 fixes both on main:
Your next push re-runs validation against main. A production-stack run with One change is needed before this can merge, though: There are two ways to resolve it: remove the setting, or move the entry to |
|
Update: I merged main into the branch (57fd21a), so validation is running again now with the fixed validate.sh and authsvc. Pull before you push the lendBodies change. |
…handling lendBodies is off by default in MoroJS and no deployment guide recommends it, so it does not belong in a Standard mode entry. 8gbit needs a re-run: the saved result was measured with it on.
|
/benchmark -f morojs -t 8gbit,production-stack --save |
|
👋 Benchmark request received. A collaborator will review and approve the run. |
Benchmark ResultsFramework:
Composite score, main against this run, in the board's default view:
The results on main were measured in another session, so a few points up or down can be the machine and not the change. Full log |
|
Hmm, did you change engine? Not sure this is emerging standard still then, I'll take a deeper look tommorow |
|
No engine change: it's the same @morojs/engine the entry has declared since #1678, and it's a regular dependency of the framework and its default server. What changed is the release. Until now the engine only spoke HTTP/1.1, so the entry forced |
Updates the morojs entry to the latest framework and engine releases.
server.lendBodiesis enabled for the echo route.Details are in the commit message and the entry README.