fix: increase file descriptor limit in parallel test run (agntcy#1672)

# Description

Local data-plane tests can fail because of exceeding the concurrent file
descriptor limit of the process (on MacOS it is 256 by default).


# Actions
- setting  fd limit to 4096 (`ulimit -n 4096`)

## Type of Change

- [x] Bugfix
- [ ] New Feature
- [ ] Breaking Change
- [ ] Refactor
- [ ] Documentation
- [ ] Other (please describe)

## Checklist

- [x] I have read the [contributing
guidelines](/agntcy/repo-template/blob/main/CONTRIBUTING.md)
- [x] Existing issues have been referenced (where applicable)
- [x] I have verified this change is not present in other open pull
requests
- [x] Functionality is documented
- [x] All code style checks pass
- [x] New code contribution is covered by automated tests
- [x] All new and existing tests pass

Signed-off-by: Mark Marton <mark.p.marton@gmail.com>

Author: markpmarton

data-plane/Cargo.lock

@@ -120,6 +120,7 @@ rand = "0.9.0"
 regex = "1.11.1"
 
 reqwest = { version = "0.12", features = ["json", "rustls-tls-no-provider", "charset", "http2", "system-proxy", "blocking"], default-features = false }
+rlimit = "0.11.0"
 rustls = { version = "0.23.31" }
 rustls-native-certs = "0.8"
 rustls-pki-types = "1.10.0"

data-plane/Cargo.toml

@@ -38,6 +38,7 @@ libsqlite3-sys = { workspace = true }
 parking_lot = { workspace = true }
 prost = { workspace = true }
 prost-types = { workspace = true }
+rlimit = { workspace = true }
 serde = { workspace = true, features = ["derive"] }
 serde_json = { workspace = true }
 serde_yaml = { workspace = true }
@@ -63,6 +64,7 @@ agntcy-slim-service = { workspace = true, features = ["session"] }
 agntcy-slim-testing = { workspace = true }
 agntcy-slim-tracing = { workspace = true }
 drain = { workspace = true }
+libc = "0.2"
 tempfile = { workspace = true }
 tokio-test = { workspace = true }
 tonic = { workspace = true }

data-plane/control-plane/Cargo.toml

@@ -1,6 +1,7 @@
 // Copyright AGNTCY Contributors (https://github.com/agntcy)
 // SPDX-License-Identifier: Apache-2.0
 
+use rlimit::increase_nofile_limit;
 use std::net::TcpListener;
 use std::time::Duration;
 use uuid::Uuid;
@@ -29,7 +30,15 @@ use tracing_subscriber::EnvFilter;
 
 // --- Helpers ---
 
+fn raise_fd_limit() {
+    static INIT_FD_LIMIT: std::sync::Once = std::sync::Once::new();
+    INIT_FD_LIMIT.call_once(|| {
+        let _ = increase_nofile_limit(4096).expect("unable to raise open file descriptor limit");
+    });
+}
+
 fn reserve_port() -> u16 {
+    raise_fd_limit();
     let listener = TcpListener::bind("127.0.0.1:0").expect("failed to bind test port");
     let port = listener.local_addr().expect("failed to read port").port();
     drop(listener);

data-plane/control-plane/tests/integration_test.rs

@@ -186,14 +186,14 @@ tasks:
     deps:
       - task: install-target
     cmds:
-      - cargo test --no-run {{.RELEASE}} {{.TARGET_ARGS}} {{.GLOBAL_ARGS}} {{.ARGS}}
+      - ulimit -n 4096 2>/dev/null || true; cargo test --no-run {{.RELEASE}} {{.TARGET_ARGS}} {{.GLOBAL_ARGS}} {{.ARGS}}
     vars:
       ARGS: '{{.ARGS | default "--all-targets --locked --all-features"}}'
 
   test:
     desc: "Run the tests"
     cmds:
-      - cargo test {{.RELEASE}} {{.TARGET_ARGS}} {{.GLOBAL_ARGS}} {{.ARGS}}
+      - ulimit -n 4096 2>/dev/null || true; cargo test {{.RELEASE}} {{.TARGET_ARGS}} {{.GLOBAL_ARGS}} {{.ARGS}}
       - task: bindings:test
       - task: core:test
     vars:
@@ -206,7 +206,7 @@ tasks:
         vars:
           COMPONENTS: "cargo-llvm-cov"
     cmds:
-      - cargo llvm-cov --lcov --output-path lcov.info {{.GLOBAL_ARGS}} {{.ARGS}}
+      - ulimit -n 4096 2>/dev/null || true; cargo llvm-cov --lcov --output-path lcov.info {{.GLOBAL_ARGS}} {{.ARGS}}
     vars:
       ARGS: '{{.ARGS | default "--all-targets --locked --all-features"}}'