refactor: prometheus upgrade (#452)

Cargo.lock

@@ -160,9 +160,9 @@ dependencies = [
 
 [[package]]
 name = "async-nats"
-version = "0.38.0"
+version = "0.40.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "76433c4de73442daedb3a59e991d94e85c14ebfc33db53dfcd347a21cd6ef4f8"
+checksum = "e23419d455dc57d3ae60a2f4278cf561fc74fe866e548e14d2b0ad3e1b8ca0b2"
 dependencies = [
  "base64 0.22.1",
  "bytes",
@@ -3481,20 +3481,17 @@ dependencies = [
 name = "metrics"
 version = "0.1.0"
 dependencies = [
- "async-nats",
  "axum 0.6.20",
  "clap",
  "dynamo-llm",
  "dynamo-runtime",
  "futures",
- "opentelemetry",
- "opentelemetry-prometheus",
  "prometheus",
  "rand 0.9.0",
  "reqwest",
  "serde",
  "serde_json",
- "thiserror 1.0.69",
+ "thiserror 2.0.12",
  "tokio",
  "tracing",
 ]
@@ -4191,80 +4188,12 @@ version = "0.1.6"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "d05e27ee213611ffe7d6348b942e8f942b37114c00cc03cec254295a4a17852e"
 
-[[package]]
-name = "opentelemetry"
-version = "0.20.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "9591d937bc0e6d2feb6f71a559540ab300ea49955229c347a517a28d27784c54"
-dependencies = [
- "opentelemetry_api",
- "opentelemetry_sdk",
-]
-
-[[package]]
-name = "opentelemetry-prometheus"
-version = "0.13.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "c7d81bc254e2d572120363a2b16cdb0d715d301b5789be0cfc26ad87e4e10e53"
-dependencies = [
- "once_cell",
- "opentelemetry_api",
- "opentelemetry_sdk",
- "prometheus",
- "protobuf",
-]
-
-[[package]]
-name = "opentelemetry_api"
-version = "0.20.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "8a81f725323db1b1206ca3da8bb19874bbd3f57c3bcd59471bfb04525b265b9b"
-dependencies = [
- "futures-channel",
- "futures-util",
- "indexmap 1.9.3",
- "js-sys",
- "once_cell",
- "pin-project-lite",
- "thiserror 1.0.69",
- "urlencoding",
-]
-
-[[package]]
-name = "opentelemetry_sdk"
-version = "0.20.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "fa8e705a0612d48139799fcbaba0d4a90f06277153e43dd2bdc16c6f0edd8026"
-dependencies = [
- "async-trait",
- "crossbeam-channel",
- "futures-channel",
- "futures-executor",
- "futures-util",
- "once_cell",
- "opentelemetry_api",
- "ordered-float",
- "percent-encoding",
- "rand 0.8.5",
- "regex",
- "thiserror 1.0.69",
-]
-
 [[package]]
 name = "option-ext"
 version = "0.2.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "04744f49eae99ab78e0d5c0b603ab218f515ea8cfe5a456d7629ad883a3b6e7d"
 
-[[package]]
-name = "ordered-float"
-version = "3.9.2"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "f1e1c390732d15f1d48471625cd92d154e66db2c56645e29a9cd26f4699f72dc"
-dependencies = [
- "num-traits",
-]
-
 [[package]]
 name = "overload"
 version = "0.1.1"
@@ -4576,9 +4505,9 @@ dependencies = [
 
 [[package]]
 name = "prometheus"
-version = "0.13.4"
+version = "0.14.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "3d33c28a30771f7f96db69893f78b857f7450d7e0237e9c8fc6427a81bae7ed1"
+checksum = "3ca5326d8d0b950a9acd87e6a3f94745394f62e4dae1b1ee22b2bc0c394af43a"
 dependencies = [
  "cfg-if 1.0.0",
  "fnv",
@@ -4586,7 +4515,7 @@ dependencies = [
  "memchr",
  "parking_lot",
  "protobuf",
- "thiserror 1.0.69",
+ "thiserror 2.0.12",
 ]
 
 [[package]]
@@ -4686,9 +4615,23 @@ dependencies = [
 
 [[package]]
 name = "protobuf"
-version = "2.28.0"
+version = "3.7.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d65a1d4ddae7d8b5de68153b48f6aa3bba8cb002b243dbdbc55a5afbc98f99f4"
+dependencies = [
+ "once_cell",
+ "protobuf-support",
+ "thiserror 1.0.69",
+]
+
+[[package]]
+name = "protobuf-support"
+version = "3.7.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "106dd99e98437432fed6519dedecfade6a06a73bb7b2a1e019fdd2bee5778d94"
+checksum = "3e36c2f31e0a47f9280fb347ef5e461ffcd2c52dd520d8e216b52f93b0b0d7d6"
+dependencies = [
+ "thiserror 1.0.69",
+]
 
 [[package]]
 name = "pulp"
@@ -5428,7 +5371,6 @@ version = "0.102.8"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "64ca1bc8749bd4cf37b5ce386cc146580777b4e8572c7b97baf22c83f444bee9"
 dependencies = [
- "ring",
  "rustls-pki-types",
  "untrusted",
 ]
@@ -6417,11 +6359,11 @@ dependencies = [
  "httparse",
  "rand 0.8.5",
  "ring",
- "rustls-native-certs 0.8.1",
  "rustls-pki-types",
  "tokio",
  "tokio-rustls",
  "tokio-util",
+ "webpki-roots",
 ]
 
 [[package]]
@@ -6866,12 +6808,6 @@ dependencies = [
  "percent-encoding",
 ]
 
-[[package]]
-name = "urlencoding"
-version = "2.1.3"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "daf8dba3b7eb870caf1ddeed7bc9d2a049f3cfdfae7cb521b087cc33ae4c49da"
-
 [[package]]
 name = "utf16_iter"
 version = "1.0.5"

Cargo.toml

@@ -40,7 +40,7 @@ dynamo-llm = { path = "lib/llm" }
 
 # External dependencies
 anyhow = { version = "1" }
-async-nats = { version = "0.38", features = ["service"] }
+async-nats = { version = "0.40", features = ["service"] }
 async-stream = { version = "0.3" }
 async-trait = { version = "0.1" }
 async_zmq = { version = "0.4.0" }
@@ -53,6 +53,7 @@ either = { version = "1.13", features = ["serde"] }
 futures = { version = "0.3" }
 hf-hub = { version = "0.4.2", default-features = false, features = ["tokio", "rustls-tls"] }
 humantime = { version = "2.2.0" }
+prometheus = { version = "0.14" }
 rand = { version = "0.9.0" }
 serde = { version = "1", features = ["derive"] }
 serde_json = { version = "1" }
@@ -66,7 +67,6 @@ validator = { version = "0.20.0", features = ["derive"] }
 uuid = { version = "1", features = ["v4", "serde"] }
 xxhash-rust = { version = "0.8", features = ["xxh3", "const_xxh3"] }
 strum = { version = "0.27", features = ["derive"] }
-prometheus = { version = "0.13" }
 
 [profile.dev.package]
 insta.opt-level = 3

components/metrics/Cargo.toml

@@ -27,17 +27,15 @@ dynamo-llm = { workspace = true }
 dynamo-runtime = { workspace = true }
 
 futures = { workspace = true }
+prometheus = { workspace = true }
 rand = { workspace = true }
 serde = { workspace = true }
 serde_json = { workspace = true }
+thiserror = { workspace = true }
 tokio = { workspace = true }
 tracing = { workspace = true }
 
-async-nats = { version = "0.38", features = ["service"] }
+# TODO: Update axum to 0.8
 axum = { version = "0.6" }
 clap = { version = "4.5", features = ["derive", "env"] }
-opentelemetry = { version = "0.20" }
-opentelemetry-prometheus = { version = "0.13" }
-prometheus = { version = "0.13" }
 reqwest = { version = "0.12", default-features = false, features = ["json", "rustls-tls"] }
-thiserror = { version = "1.0" }

components/metrics/src/bin/mock_worker.rs

@@ -13,12 +13,11 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
-use async_nats::service::endpoint::Stats;
 use dynamo_llm::kv_router::{
     protocols::ForwardPassMetrics, scheduler::KVHitRateEvent, KV_HIT_RATE_SUBJECT,
 };
 use dynamo_runtime::{
-    component::Namespace,
+    component::{service::EndpointStats, Namespace},
     logging,
     pipeline::{
         async_trait, network::Ingress, AsyncEngine, AsyncEngineContextProvider, Error, ManyOut,
@@ -107,7 +106,7 @@ async fn mock_event_publisher(namespace: Namespace) {
 }
 
 /// Generates mock forward pass metrics for stats handler
-fn mock_stats_handler(_stats: Stats) -> serde_json::Value {
+fn mock_stats_handler(_stats: EndpointStats) -> serde_json::Value {
     let request_total_slots = 100;
     let request_active_slots = rand::rng().random_range(0..=request_total_slots);
     let kv_total_blocks = 100;

components/metrics/src/lib.rs

@@ -429,7 +429,7 @@ impl PrometheusMetrics {
         &self,
         gauge: &prometheus::GaugeVec,
         config: &LLMWorkerLoadCapacityConfig,
-        worker_id: &str,
+        worker_id: &String,
         value: f64,
     ) {
         gauge
@@ -442,7 +442,7 @@ impl PrometheusMetrics {
         &self,
         counter: &prometheus::CounterVec,
         config: &LLMWorkerLoadCapacityConfig,
-        worker_id: &str,
+        worker_id: &String,
         value: f64,
     ) {
         counter

lib/bindings/python/Cargo.lock

@@ -138,9 +138,9 @@ dependencies = [
 
 [[package]]
 name = "async-nats"
-version = "0.38.0"
+version = "0.40.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "76433c4de73442daedb3a59e991d94e85c14ebfc33db53dfcd347a21cd6ef4f8"
+checksum = "e23419d455dc57d3ae60a2f4278cf561fc74fe866e548e14d2b0ad3e1b8ca0b2"
 dependencies = [
  "base64 0.22.1",
  "bytes",
@@ -3144,9 +3144,9 @@ dependencies = [
 
 [[package]]
 name = "prometheus"
-version = "0.13.4"
+version = "0.14.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "3d33c28a30771f7f96db69893f78b857f7450d7e0237e9c8fc6427a81bae7ed1"
+checksum = "3ca5326d8d0b950a9acd87e6a3f94745394f62e4dae1b1ee22b2bc0c394af43a"
 dependencies = [
  "cfg-if 1.0.0",
  "fnv",
@@ -3154,7 +3154,7 @@ dependencies = [
  "memchr",
  "parking_lot",
  "protobuf",
- "thiserror 1.0.69",
+ "thiserror 2.0.12",
 ]
 
 [[package]]
@@ -3211,9 +3211,23 @@ dependencies = [
 
 [[package]]
 name = "protobuf"
-version = "2.28.0"
+version = "3.7.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d65a1d4ddae7d8b5de68153b48f6aa3bba8cb002b243dbdbc55a5afbc98f99f4"
+dependencies = [
+ "once_cell",
+ "protobuf-support",
+ "thiserror 1.0.69",
+]
+
+[[package]]
+name = "protobuf-support"
+version = "3.7.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "106dd99e98437432fed6519dedecfade6a06a73bb7b2a1e019fdd2bee5778d94"
+checksum = "3e36c2f31e0a47f9280fb347ef5e461ffcd2c52dd520d8e216b52f93b0b0d7d6"
+dependencies = [
+ "thiserror 1.0.69",
+]
 
 [[package]]
 name = "pulp"
@@ -3781,7 +3795,6 @@ version = "0.102.8"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "64ca1bc8749bd4cf37b5ce386cc146580777b4e8572c7b97baf22c83f444bee9"
 dependencies = [
- "ring",
  "rustls-pki-types",
  "untrusted",
 ]
@@ -4481,11 +4494,11 @@ dependencies = [
  "httparse",
  "rand 0.8.5",
  "ring",
- "rustls-native-certs 0.8.1",
  "rustls-pki-types",
  "tokio",
  "tokio-rustls",
  "tokio-util",
+ "webpki-roots",
 ]
 
 [[package]]

lib/runtime/examples/Cargo.lock

@@ -61,9 +61,9 @@ checksum = "7c02d123df017efcdfbd739ef81735b36c5ba83ec3c59c80a9d7ecc718f92e50"
 
 [[package]]
 name = "async-nats"
-version = "0.38.0"
+version = "0.40.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "76433c4de73442daedb3a59e991d94e85c14ebfc33db53dfcd347a21cd6ef4f8"
+checksum = "e23419d455dc57d3ae60a2f4278cf561fc74fe866e548e14d2b0ad3e1b8ca0b2"
 dependencies = [
  "base64",
  "bytes",
@@ -77,7 +77,7 @@ dependencies = [
  "rand 0.8.5",
  "regex",
  "ring",
- "rustls-native-certs 0.7.3",
+ "rustls-native-certs",
  "rustls-pemfile",
  "rustls-webpki 0.102.8",
  "serde",
@@ -378,16 +378,6 @@ dependencies = [
  "libc",
 ]
 
-[[package]]
-name = "core-foundation"
-version = "0.10.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "b55271e5c8c478ad3f38ad24ef34923091e0548492a266d19b3c0b4d82574c63"
-dependencies = [
- "core-foundation-sys",
- "libc",
-]
-
 [[package]]
 name = "core-foundation-sys"
 version = "0.8.7"
@@ -1846,9 +1836,9 @@ dependencies = [
 
 [[package]]
 name = "prometheus"
-version = "0.13.4"
+version = "0.14.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "3d33c28a30771f7f96db69893f78b857f7450d7e0237e9c8fc6427a81bae7ed1"
+checksum = "3ca5326d8d0b950a9acd87e6a3f94745394f62e4dae1b1ee22b2bc0c394af43a"
 dependencies = [
  "cfg-if 1.0.0",
  "fnv",
@@ -1856,7 +1846,7 @@ dependencies = [
  "memchr",
  "parking_lot",
  "protobuf",
- "thiserror 1.0.69",
+ "thiserror 2.0.12",
 ]
 
 [[package]]
@@ -1913,9 +1903,23 @@ dependencies = [
 
 [[package]]
 name = "protobuf"
-version = "2.28.0"
+version = "3.7.2"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "106dd99e98437432fed6519dedecfade6a06a73bb7b2a1e019fdd2bee5778d94"
+checksum = "d65a1d4ddae7d8b5de68153b48f6aa3bba8cb002b243dbdbc55a5afbc98f99f4"
+dependencies = [
+ "once_cell",
+ "protobuf-support",
+ "thiserror 1.0.69",
+]
+
+[[package]]
+name = "protobuf-support"
+version = "3.7.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "3e36c2f31e0a47f9280fb347ef5e461ffcd2c52dd520d8e216b52f93b0b0d7d6"
+dependencies = [
+ "thiserror 1.0.69",
+]
 
 [[package]]
 name = "quote"
@@ -2131,19 +2135,7 @@ dependencies = [
  "rustls-pemfile",
  "rustls-pki-types",
  "schannel",
- "security-framework 2.11.1",
-]
-
-[[package]]
-name = "rustls-native-certs"
-version = "0.8.1"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "7fcff2dd52b58a8d98a70243663a0d234c4e2b79235637849d15913394a247d3"
-dependencies = [
- "openssl-probe",
- "rustls-pki-types",
- "schannel",
- "security-framework 3.2.0",
+ "security-framework",
 ]
 
 [[package]]
@@ -2167,7 +2159,6 @@ version = "0.102.8"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "64ca1bc8749bd4cf37b5ce386cc146580777b4e8572c7b97baf22c83f444bee9"
 dependencies = [
- "ring",
  "rustls-pki-types",
  "untrusted",
 ]
@@ -2226,20 +2217,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "897b2245f0b511c87893af39b033e5ca9cce68824c4d7e7630b5a1d339658d02"
 dependencies = [
  "bitflags 2.9.0",
- "core-foundation 0.9.4",
- "core-foundation-sys",
- "libc",
- "security-framework-sys",
-]
-
-[[package]]
-name = "security-framework"
-version = "3.2.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "271720403f46ca04f7ba6f55d438f8bd878d6b8ca0a1046e8228c4145bcbb316"
-dependencies = [
- "bitflags 2.9.0",
- "core-foundation 0.10.0",
+ "core-foundation",
  "core-foundation-sys",
  "libc",
  "security-framework-sys",
@@ -2684,11 +2662,11 @@ dependencies = [
  "httparse",
  "rand 0.8.5",
  "ring",
- "rustls-native-certs 0.8.1",
  "rustls-pki-types",
  "tokio",
  "tokio-rustls",
  "tokio-util",
+ "webpki-roots",
 ]
 
 [[package]]
@@ -3107,6 +3085,15 @@ dependencies = [
  "unicode-ident",
 ]
 
+[[package]]
+name = "webpki-roots"
+version = "0.26.8"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "2210b291f7ea53617fbafcc4939f10914214ec15aace5ba62293a668f322c5c9"
+dependencies = [
+ "rustls-pki-types",
+]
+
 [[package]]
 name = "winapi"
 version = "0.2.8"

lib/runtime/src/component.rs

@@ -66,7 +66,7 @@ mod component;
 mod endpoint;
 mod namespace;
 mod registry;
-mod service;
+pub mod service;
 
 pub use client::Client;
 

lib/runtime/src/component/endpoint.rs

@@ -17,6 +17,8 @@ use derive_getters::Dissolve;
 
 use super::*;
 
+pub use async_nats::service::endpoint::Stats as EndpointStats;
+
 #[derive(Educe, Builder, Dissolve)]
 #[educe(Debug)]
 #[builder(pattern = "owned", build_fn(private, name = "build_internal"))]
@@ -47,7 +49,7 @@ impl EndpointConfigBuilder {
 
     pub fn stats_handler<F>(self, handler: F) -> Self
     where
-        F: FnMut(async_nats::service::endpoint::Stats) -> serde_json::Value + Send + Sync + 'static,
+        F: FnMut(EndpointStats) -> serde_json::Value + Send + Sync + 'static,
     {
         self._stats_handler(Some(Box::new(handler)))
     }

lib/runtime/src/component/service.rs

@@ -19,13 +19,11 @@ use std::sync::Mutex;
 
 use super::*;
 
-use async_nats::service::{endpoint, Service};
-
+pub use super::endpoint::EndpointStats;
 pub type StatsHandler =
-    Box<dyn FnMut(String, endpoint::Stats) -> serde_json::Value + Send + Sync + 'static>;
-
+    Box<dyn FnMut(String, EndpointStats) -> serde_json::Value + Send + Sync + 'static>;
 pub type EndpointStatsHandler =
-    Box<dyn FnMut(endpoint::Stats) -> serde_json::Value + Send + Sync + 'static>;
+    Box<dyn FnMut(EndpointStats) -> serde_json::Value + Send + Sync + 'static>;
 
 pub const PROJECT_NAME: &str = "Dynamo";