fix(infra): upgrade OpenTelemetry 0.27→0.31 to align tonic versions, fix DCGM scheduling
OTLP fix: opentelemetry-otlp 0.27 bundled tonic 0.12 while the workspace uses tonic 0.14, making with_channel() impossible (type mismatch). Upgrading to otel-otlp 0.31 aligns both on tonic 0.14, enabling explicit Channel construction that respects http:// scheme (no spurious TLS negotiation). API migrations (otel 0.27→0.31): - TracerProvider → SdkTracerProvider - with_batch_exporter(exporter, runtime) → with_batch_exporter(exporter) - Resource::new(vec![...]) → Resource::builder().with_service_name().build() - global::shutdown_tracer_provider() removed (Drop-based shutdown) - opentelemetry-otlp feature "tonic" → "grpc-tonic" DCGM fix: remove runtimeClassName: nvidia from DaemonSet — Scaleway Kapsule GPU pools use nvidia runtime as default containerd handler. The RuntimeClass CRD is only created by the full GPU Operator, not the device plugin alone. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
This commit is contained in:
jgrusewski
187
Cargo.lock
generated
187
Cargo.lock
generated
@@ -226,7 +226,7 @@ dependencies = [
|
||||
"tokio",
|
||||
"tokio-stream",
|
||||
"tokio-test",
|
||||
"tonic 0.14.2",
|
||||
"tonic",
|
||||
"tonic-health",
|
||||
"tonic-prost",
|
||||
"tonic-prost-build",
|
||||
@@ -270,7 +270,7 @@ dependencies = [
|
||||
"tokio",
|
||||
"tokio-stream",
|
||||
"toml",
|
||||
"tonic 0.14.2",
|
||||
"tonic",
|
||||
"tonic-prost",
|
||||
"tracing",
|
||||
"tracing-subscriber",
|
||||
@@ -610,7 +610,7 @@ dependencies = [
|
||||
"arrow-schema 55.2.0",
|
||||
"chrono",
|
||||
"half",
|
||||
"indexmap 2.11.4",
|
||||
"indexmap",
|
||||
"lexical-core",
|
||||
"memchr",
|
||||
"num",
|
||||
@@ -1565,7 +1565,7 @@ dependencies = [
|
||||
"tokio-stream",
|
||||
"tokio-test",
|
||||
"tokio-util",
|
||||
"tonic 0.14.2",
|
||||
"tonic",
|
||||
"tonic-health",
|
||||
"tonic-prost",
|
||||
"tonic-prost-build",
|
||||
@@ -1824,7 +1824,7 @@ dependencies = [
|
||||
"tokio",
|
||||
"tokio-stream",
|
||||
"tokio-test",
|
||||
"tonic 0.14.2",
|
||||
"tonic",
|
||||
"tonic-health",
|
||||
"tonic-prost",
|
||||
"tonic-prost-build",
|
||||
@@ -2313,7 +2313,7 @@ dependencies = [
|
||||
"tokio",
|
||||
"tokio-test",
|
||||
"toml",
|
||||
"tonic 0.14.2",
|
||||
"tonic",
|
||||
"tower-layer",
|
||||
"tower-service",
|
||||
"tracing",
|
||||
@@ -2985,7 +2985,7 @@ dependencies = [
|
||||
"tokio-retry",
|
||||
"tokio-stream",
|
||||
"tokio-util",
|
||||
"tonic 0.14.2",
|
||||
"tonic",
|
||||
"tonic-prost",
|
||||
"tonic-prost-build",
|
||||
"tonic-reflection",
|
||||
@@ -3841,7 +3841,7 @@ dependencies = [
|
||||
"thiserror 1.0.69",
|
||||
"tokio",
|
||||
"tokio-stream",
|
||||
"tonic 0.14.2",
|
||||
"tonic",
|
||||
"tracing",
|
||||
"tracing-subscriber",
|
||||
"trading_engine",
|
||||
@@ -3882,7 +3882,7 @@ dependencies = [
|
||||
"tokio",
|
||||
"tokio-stream",
|
||||
"tokio-test",
|
||||
"tonic 0.14.2",
|
||||
"tonic",
|
||||
"tonic-prost",
|
||||
"tonic-prost-build",
|
||||
"tracing",
|
||||
@@ -4090,7 +4090,7 @@ dependencies = [
|
||||
"tokio-stream",
|
||||
"tokio-test",
|
||||
"toml",
|
||||
"tonic 0.14.2",
|
||||
"tonic",
|
||||
"tonic-prost",
|
||||
"tonic-prost-build",
|
||||
"tracing",
|
||||
@@ -4358,7 +4358,7 @@ dependencies = [
|
||||
"futures-sink",
|
||||
"futures-util",
|
||||
"http 0.2.12",
|
||||
"indexmap 2.11.4",
|
||||
"indexmap",
|
||||
"slab",
|
||||
"tokio",
|
||||
"tokio-util",
|
||||
@@ -4377,7 +4377,7 @@ dependencies = [
|
||||
"futures-core",
|
||||
"futures-sink",
|
||||
"http 1.3.1",
|
||||
"indexmap 2.11.4",
|
||||
"indexmap",
|
||||
"slab",
|
||||
"tokio",
|
||||
"tokio-util",
|
||||
@@ -5002,16 +5002,6 @@ version = "1.12.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "e7c5cedc30da3a610cac6b4ba17597bdf7152cf974e8aab3afb3d54455e371c8"
|
||||
|
||||
[[package]]
|
||||
name = "indexmap"
|
||||
version = "1.9.3"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "bd070e393353796e801d209ad339e89596eb4c8d430d18ede6a1cced8fafbd99"
|
||||
dependencies = [
|
||||
"autocfg",
|
||||
"hashbrown 0.12.3",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "indexmap"
|
||||
version = "2.11.4"
|
||||
@@ -5155,7 +5145,7 @@ dependencies = [
|
||||
"prost-build",
|
||||
"reqwest 0.12.23",
|
||||
"tokio",
|
||||
"tonic 0.14.2",
|
||||
"tonic",
|
||||
"tonic-prost",
|
||||
"tonic-prost-build",
|
||||
"uuid",
|
||||
@@ -5181,7 +5171,7 @@ dependencies = [
|
||||
"serial_test",
|
||||
"thiserror 1.0.69",
|
||||
"tokio",
|
||||
"tonic 0.14.2",
|
||||
"tonic",
|
||||
"tonic-prost",
|
||||
"tonic-prost-build",
|
||||
"tracing",
|
||||
@@ -5863,7 +5853,7 @@ dependencies = [
|
||||
"hyper 1.7.0",
|
||||
"hyper-rustls 0.27.7",
|
||||
"hyper-util",
|
||||
"indexmap 2.11.4",
|
||||
"indexmap",
|
||||
"ipnet",
|
||||
"metrics",
|
||||
"metrics-util",
|
||||
@@ -6108,7 +6098,7 @@ dependencies = [
|
||||
"tokio-retry",
|
||||
"tokio-stream",
|
||||
"tokio-util",
|
||||
"tonic 0.14.2",
|
||||
"tonic",
|
||||
"tonic-prost",
|
||||
"tonic-prost-build",
|
||||
"tonic-reflection",
|
||||
@@ -6651,68 +6641,76 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "opentelemetry"
|
||||
version = "0.27.1"
|
||||
version = "0.31.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "ab70038c28ed37b97d8ed414b6429d343a8bbf44c9f79ec854f3a643029ba6d7"
|
||||
checksum = "b84bcd6ae87133e903af7ef497404dda70c60d0ea14895fc8a5e6722754fc2a0"
|
||||
dependencies = [
|
||||
"futures-core",
|
||||
"futures-sink",
|
||||
"js-sys",
|
||||
"pin-project-lite",
|
||||
"thiserror 1.0.69",
|
||||
"thiserror 2.0.17",
|
||||
"tracing",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "opentelemetry-otlp"
|
||||
version = "0.27.0"
|
||||
name = "opentelemetry-http"
|
||||
version = "0.31.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "91cf61a1868dacc576bf2b2a1c3e9ab150af7272909e80085c3173384fe11f76"
|
||||
checksum = "d7a6d09a73194e6b66df7c8f1b680f156d916a1a942abf2de06823dd02b7855d"
|
||||
dependencies = [
|
||||
"async-trait",
|
||||
"futures-core",
|
||||
"bytes",
|
||||
"http 1.3.1",
|
||||
"opentelemetry",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "opentelemetry-otlp"
|
||||
version = "0.31.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "7a2366db2dca4d2ad033cad11e6ee42844fd727007af5ad04a1730f4cb8163bf"
|
||||
dependencies = [
|
||||
"http 1.3.1",
|
||||
"opentelemetry",
|
||||
"opentelemetry-http",
|
||||
"opentelemetry-proto",
|
||||
"opentelemetry_sdk",
|
||||
"prost 0.13.5",
|
||||
"thiserror 1.0.69",
|
||||
"prost 0.14.1",
|
||||
"thiserror 2.0.17",
|
||||
"tokio",
|
||||
"tonic 0.12.3",
|
||||
"tonic",
|
||||
"tracing",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "opentelemetry-proto"
|
||||
version = "0.27.0"
|
||||
version = "0.31.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "a6e05acbfada5ec79023c85368af14abd0b307c015e9064d249b2a950ef459a6"
|
||||
checksum = "a7175df06de5eaee9909d4805a3d07e28bb752c34cab57fa9cff549da596b30f"
|
||||
dependencies = [
|
||||
"opentelemetry",
|
||||
"opentelemetry_sdk",
|
||||
"prost 0.13.5",
|
||||
"tonic 0.12.3",
|
||||
"prost 0.14.1",
|
||||
"tonic",
|
||||
"tonic-prost",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "opentelemetry_sdk"
|
||||
version = "0.27.1"
|
||||
version = "0.31.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "231e9d6ceef9b0b2546ddf52335785ce41252bc7474ee8ba05bfad277be13ab8"
|
||||
checksum = "e14ae4f5991976fd48df6d843de219ca6d31b01daaab2dad5af2badeded372bd"
|
||||
dependencies = [
|
||||
"async-trait",
|
||||
"futures-channel",
|
||||
"futures-executor",
|
||||
"futures-util",
|
||||
"glob",
|
||||
"opentelemetry",
|
||||
"percent-encoding",
|
||||
"rand 0.8.5",
|
||||
"serde_json",
|
||||
"thiserror 1.0.69",
|
||||
"rand 0.9.2",
|
||||
"thiserror 2.0.17",
|
||||
"tokio",
|
||||
"tokio-stream",
|
||||
"tracing",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
@@ -7014,7 +7012,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "b4c5cc86750666a3ed20bdaf5ca2a0344f9c67674cae0515bec2da16fbaa47db"
|
||||
dependencies = [
|
||||
"fixedbitset 0.4.2",
|
||||
"indexmap 2.11.4",
|
||||
"indexmap",
|
||||
"serde",
|
||||
]
|
||||
|
||||
@@ -7025,7 +7023,7 @@ source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "3672b37090dbd86368a4145bc067582552b29c27377cad4e0a306c97f9bd7772"
|
||||
dependencies = [
|
||||
"fixedbitset 0.5.7",
|
||||
"indexmap 2.11.4",
|
||||
"indexmap",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
@@ -9034,7 +9032,7 @@ version = "0.9.34+deprecated"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "6a8b1a1a2ebf674015cc02edccce75287f1a0130d394307b36743c2f5d504b47"
|
||||
dependencies = [
|
||||
"indexmap 2.11.4",
|
||||
"indexmap",
|
||||
"itoa",
|
||||
"ryu",
|
||||
"serde",
|
||||
@@ -9428,7 +9426,7 @@ dependencies = [
|
||||
"futures-util",
|
||||
"hashbrown 0.15.5",
|
||||
"hashlink",
|
||||
"indexmap 2.11.4",
|
||||
"indexmap",
|
||||
"log",
|
||||
"memchr",
|
||||
"once_cell",
|
||||
@@ -9641,7 +9639,7 @@ dependencies = [
|
||||
"flate2",
|
||||
"fs2",
|
||||
"futures",
|
||||
"indexmap 2.11.4",
|
||||
"indexmap",
|
||||
"lru",
|
||||
"object_store",
|
||||
"parking_lot 0.12.5",
|
||||
@@ -9684,7 +9682,7 @@ dependencies = [
|
||||
"tempfile",
|
||||
"thiserror 1.0.69",
|
||||
"tokio",
|
||||
"tonic 0.14.2",
|
||||
"tonic",
|
||||
"tonic-prost",
|
||||
"tracing",
|
||||
"tracing-subscriber",
|
||||
@@ -10069,7 +10067,7 @@ dependencies = [
|
||||
"tokio-stream",
|
||||
"tokio-test",
|
||||
"toml",
|
||||
"tonic 0.14.2",
|
||||
"tonic",
|
||||
"tonic-health",
|
||||
"tracing",
|
||||
"tracing-subscriber",
|
||||
@@ -10437,7 +10435,7 @@ version = "0.22.27"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "41fe8c660ae4257887cf66394862d21dbca4a6ddd26f04a3560410406a2f819a"
|
||||
dependencies = [
|
||||
"indexmap 2.11.4",
|
||||
"indexmap",
|
||||
"serde",
|
||||
"serde_spanned",
|
||||
"toml_datetime 0.6.11",
|
||||
@@ -10451,7 +10449,7 @@ version = "0.23.6"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "f3effe7c0e86fdff4f69cdd2ccc1b96f933e24811c5441d44904e8683e27184b"
|
||||
dependencies = [
|
||||
"indexmap 2.11.4",
|
||||
"indexmap",
|
||||
"toml_datetime 0.7.2",
|
||||
"toml_parser",
|
||||
"winnow",
|
||||
@@ -10472,36 +10470,6 @@ version = "0.1.2"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "5d99f8c9a7727884afe522e9bd5edbfc91a3312b36a77b5fb8926e4c31a41801"
|
||||
|
||||
[[package]]
|
||||
name = "tonic"
|
||||
version = "0.12.3"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "877c5b330756d856ffcc4553ab34a5684481ade925ecc54bcd1bf02b1d0d4d52"
|
||||
dependencies = [
|
||||
"async-stream",
|
||||
"async-trait",
|
||||
"axum 0.7.9",
|
||||
"base64 0.22.1",
|
||||
"bytes",
|
||||
"h2 0.4.12",
|
||||
"http 1.3.1",
|
||||
"http-body 1.0.1",
|
||||
"http-body-util",
|
||||
"hyper 1.7.0",
|
||||
"hyper-timeout",
|
||||
"hyper-util",
|
||||
"percent-encoding",
|
||||
"pin-project",
|
||||
"prost 0.13.5",
|
||||
"socket2 0.5.10",
|
||||
"tokio",
|
||||
"tokio-stream",
|
||||
"tower 0.4.13",
|
||||
"tower-layer",
|
||||
"tower-service",
|
||||
"tracing",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "tonic"
|
||||
version = "0.14.2"
|
||||
@@ -10554,7 +10522,7 @@ dependencies = [
|
||||
"prost 0.14.1",
|
||||
"tokio",
|
||||
"tokio-stream",
|
||||
"tonic 0.14.2",
|
||||
"tonic",
|
||||
"tonic-prost",
|
||||
]
|
||||
|
||||
@@ -10566,7 +10534,7 @@ checksum = "66bd50ad6ce1252d87ef024b3d64fe4c3cf54a86fb9ef4c631fdd0ded7aeaa67"
|
||||
dependencies = [
|
||||
"bytes",
|
||||
"prost 0.14.1",
|
||||
"tonic 0.14.2",
|
||||
"tonic",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
@@ -10595,7 +10563,7 @@ dependencies = [
|
||||
"prost-types",
|
||||
"tokio",
|
||||
"tokio-stream",
|
||||
"tonic 0.14.2",
|
||||
"tonic",
|
||||
"tonic-prost",
|
||||
]
|
||||
|
||||
@@ -10620,11 +10588,8 @@ checksum = "b8fa9be0de6cf49e536ce1851f987bd21a43b771b09473c3549a6c853db37c1c"
|
||||
dependencies = [
|
||||
"futures-core",
|
||||
"futures-util",
|
||||
"indexmap 1.9.3",
|
||||
"pin-project",
|
||||
"pin-project-lite",
|
||||
"rand 0.8.5",
|
||||
"slab",
|
||||
"tokio",
|
||||
"tokio-util",
|
||||
"tower-layer",
|
||||
@@ -10640,7 +10605,7 @@ checksum = "d039ad9159c98b70ecfd540b2573b97f7f52c3e8d9f8ad57a24b916a536975f9"
|
||||
dependencies = [
|
||||
"futures-core",
|
||||
"futures-util",
|
||||
"indexmap 2.11.4",
|
||||
"indexmap",
|
||||
"pin-project-lite",
|
||||
"slab",
|
||||
"sync_wrapper 1.0.2",
|
||||
@@ -10717,9 +10682,9 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "tracing"
|
||||
version = "0.1.41"
|
||||
version = "0.1.44"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "784e0ac535deb450455cbfa28a6f0df145ea1bb7ae51b821cf5e7927fdcfbdd0"
|
||||
checksum = "63e71662fa4b2a2c3a26f570f037eb95bb1f85397f3cd8076caed2f026a6d100"
|
||||
dependencies = [
|
||||
"log",
|
||||
"pin-project-lite",
|
||||
@@ -10741,9 +10706,9 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "tracing-attributes"
|
||||
version = "0.1.30"
|
||||
version = "0.1.31"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "81383ab64e72a7a8b8e13130c49e3dab29def6d0c7d76a03087b3cf71c5c6903"
|
||||
checksum = "7490cfa5ec963746568740651ac6781f701c9c5ea257c58e057f3ba8cf69e8da"
|
||||
dependencies = [
|
||||
"proc-macro2",
|
||||
"quote",
|
||||
@@ -10752,9 +10717,9 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "tracing-core"
|
||||
version = "0.1.34"
|
||||
version = "0.1.36"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "b9d12581f227e93f094d3af2ae690a574abb8a2b9b7a96e7cfe9647b2b617678"
|
||||
checksum = "db97caf9d906fbde555dd62fa95ddba9eecfd14cb388e4f491a66d74cd5fb79a"
|
||||
dependencies = [
|
||||
"once_cell",
|
||||
"valuable",
|
||||
@@ -10783,14 +10748,12 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "tracing-opentelemetry"
|
||||
version = "0.28.0"
|
||||
version = "0.32.1"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "97a971f6058498b5c0f1affa23e7ea202057a7301dbff68e968b2d578bcbd053"
|
||||
checksum = "1ac28f2d093c6c477eaa76b23525478f38de514fa9aeb1285738d4b97a9552fc"
|
||||
dependencies = [
|
||||
"js-sys",
|
||||
"once_cell",
|
||||
"opentelemetry",
|
||||
"opentelemetry_sdk",
|
||||
"smallvec",
|
||||
"tracing",
|
||||
"tracing-core",
|
||||
@@ -10811,9 +10774,9 @@ dependencies = [
|
||||
|
||||
[[package]]
|
||||
name = "tracing-subscriber"
|
||||
version = "0.3.20"
|
||||
version = "0.3.22"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "2054a14f5307d601f88daf0553e1cbf472acc4f2c51afab632431cdcd72124d5"
|
||||
checksum = "2f30143827ddab0d256fd843b7a66d164e9f271cfa0dde49142c5ca0ca291f1e"
|
||||
dependencies = [
|
||||
"matchers",
|
||||
"nu-ansi-term",
|
||||
@@ -10889,7 +10852,7 @@ dependencies = [
|
||||
"thiserror 1.0.69",
|
||||
"tokio",
|
||||
"tokio-stream",
|
||||
"tonic 0.14.2",
|
||||
"tonic",
|
||||
"tonic-health",
|
||||
"tonic-prost",
|
||||
"tonic-prost-build",
|
||||
@@ -11016,7 +10979,7 @@ dependencies = [
|
||||
"tokio",
|
||||
"tokio-stream",
|
||||
"tokio-tungstenite",
|
||||
"tonic 0.14.2",
|
||||
"tonic",
|
||||
"tonic-health",
|
||||
"tonic-prost",
|
||||
"tonic-prost-build",
|
||||
@@ -11059,7 +11022,7 @@ dependencies = [
|
||||
"tokio",
|
||||
"tokio-stream",
|
||||
"toml",
|
||||
"tonic 0.14.2",
|
||||
"tonic",
|
||||
"tonic-prost",
|
||||
"tonic-prost-build",
|
||||
"tracing",
|
||||
@@ -11081,7 +11044,7 @@ dependencies = [
|
||||
"serde",
|
||||
"serde_json",
|
||||
"tokio",
|
||||
"tonic 0.14.2",
|
||||
"tonic",
|
||||
"tonic-prost",
|
||||
"tonic-prost-build",
|
||||
"tracing",
|
||||
@@ -11555,7 +11518,7 @@ dependencies = [
|
||||
"tokio",
|
||||
"tokio-stream",
|
||||
"tokio-test",
|
||||
"tonic 0.14.2",
|
||||
"tonic",
|
||||
"tonic-prost",
|
||||
"tonic-prost-build",
|
||||
"tower 0.4.13",
|
||||
@@ -12276,7 +12239,7 @@ dependencies = [
|
||||
"crc32fast",
|
||||
"crossbeam-utils",
|
||||
"displaydoc",
|
||||
"indexmap 2.11.4",
|
||||
"indexmap",
|
||||
"num_enum",
|
||||
"thiserror 1.0.69",
|
||||
]
|
||||
|
||||
10
Cargo.toml
10
Cargo.toml
@@ -336,10 +336,12 @@ hdrhistogram = "7.5"
|
||||
influxdb2 = { version = "0.5", default-features = false, features = ["native-tls"] }
|
||||
|
||||
# OpenTelemetry for production monitoring and extensive logging
|
||||
tracing-opentelemetry = "0.28"
|
||||
opentelemetry = { version = "0.27", features = ["trace", "metrics", "logs"] }
|
||||
opentelemetry-otlp = { version = "0.27", features = ["tonic", "metrics", "trace", "logs"] }
|
||||
opentelemetry_sdk = { version = "0.27", features = ["rt-tokio", "metrics", "trace", "logs"] }
|
||||
# NOTE: otel-otlp 0.31 uses tonic 0.14 (matching workspace), fixing the tonic version mismatch
|
||||
# that prevented with_channel() for explicit plaintext gRPC channels.
|
||||
tracing-opentelemetry = "0.32"
|
||||
opentelemetry = { version = "0.31", features = ["trace", "metrics", "logs"] }
|
||||
opentelemetry-otlp = { version = "0.31", default-features = false, features = ["grpc-tonic", "metrics", "trace", "logs", "internal-logs"] }
|
||||
opentelemetry_sdk = { version = "0.31", features = ["rt-tokio", "metrics", "trace", "logs"] }
|
||||
|
||||
# Additional commonly used dependencies - CONSOLIDATED
|
||||
# Build dependencies already defined above with tonic dependencies
|
||||
|
||||
@@ -22,10 +22,9 @@
|
||||
|
||||
use crate::error::{CommonError, CommonResult};
|
||||
use opentelemetry::{global, trace::TracerProvider as _, KeyValue};
|
||||
use opentelemetry_otlp::{SpanExporter, WithExportConfig};
|
||||
use opentelemetry_otlp::{SpanExporter, WithTonicConfig};
|
||||
use opentelemetry_sdk::{
|
||||
runtime::Tokio as OtelTokio,
|
||||
trace::{Sampler, TracerProvider},
|
||||
trace::{Sampler, SdkTracerProvider},
|
||||
Resource,
|
||||
};
|
||||
|
||||
@@ -83,9 +82,19 @@ pub fn build_otel_tracer(
|
||||
return None;
|
||||
}
|
||||
|
||||
// Build an explicit tonic Channel that respects the URI scheme (http vs https).
|
||||
// This prevents TLS negotiation on plaintext http:// endpoints like Tempo.
|
||||
let channel = match tonic::transport::Channel::from_shared(config.otlp_endpoint.clone()) {
|
||||
Ok(endpoint) => endpoint.connect_lazy(),
|
||||
Err(e) => {
|
||||
eprintln!("Invalid OTLP endpoint URI (non-fatal): {e}");
|
||||
return None;
|
||||
}
|
||||
};
|
||||
|
||||
let exporter = match SpanExporter::builder()
|
||||
.with_tonic()
|
||||
.with_endpoint(&config.otlp_endpoint)
|
||||
.with_channel(channel)
|
||||
.build()
|
||||
{
|
||||
Ok(e) => e,
|
||||
@@ -95,13 +104,15 @@ pub fn build_otel_tracer(
|
||||
}
|
||||
};
|
||||
|
||||
let provider = TracerProvider::builder()
|
||||
let provider = SdkTracerProvider::builder()
|
||||
.with_sampler(Sampler::AlwaysOn)
|
||||
.with_batch_exporter(exporter, OtelTokio)
|
||||
.with_resource(Resource::new(vec![
|
||||
KeyValue::new("service.name", config.service_name.clone()),
|
||||
KeyValue::new("service.version", env!("CARGO_PKG_VERSION")),
|
||||
]))
|
||||
.with_batch_exporter(exporter)
|
||||
.with_resource(
|
||||
Resource::builder()
|
||||
.with_service_name(config.service_name.clone())
|
||||
.with_attribute(KeyValue::new("service.version", env!("CARGO_PKG_VERSION")))
|
||||
.build(),
|
||||
)
|
||||
.build();
|
||||
|
||||
let tracer = provider.tracer(config.service_name.clone());
|
||||
@@ -153,21 +164,27 @@ pub async fn init_tracing(config: TracingConfig) -> CommonResult<()> {
|
||||
return Ok(());
|
||||
}
|
||||
|
||||
// Build OTLP span exporter targeting the configured collector endpoint
|
||||
// Build an explicit tonic Channel — see build_otel_tracer() comment for rationale
|
||||
let channel = tonic::transport::Channel::from_shared(config.otlp_endpoint.clone())
|
||||
.map_err(|e| CommonError::config(format!("Invalid OTLP endpoint URI: {}", e)))?
|
||||
.connect_lazy();
|
||||
|
||||
let exporter = SpanExporter::builder()
|
||||
.with_tonic()
|
||||
.with_endpoint(&config.otlp_endpoint)
|
||||
.with_channel(channel)
|
||||
.build()
|
||||
.map_err(|e| CommonError::config(format!("Failed to build OTLP exporter: {}", e)))?;
|
||||
|
||||
// Create a tracer provider with the OTLP exporter using batched async export
|
||||
let provider = TracerProvider::builder()
|
||||
// Create a tracer provider with the OTLP exporter using batched export
|
||||
let provider = SdkTracerProvider::builder()
|
||||
.with_sampler(Sampler::AlwaysOn)
|
||||
.with_batch_exporter(exporter, OtelTokio)
|
||||
.with_resource(Resource::new(vec![
|
||||
KeyValue::new("service.name", config.service_name.clone()),
|
||||
KeyValue::new("service.version", env!("CARGO_PKG_VERSION")),
|
||||
]))
|
||||
.with_batch_exporter(exporter)
|
||||
.with_resource(
|
||||
Resource::builder()
|
||||
.with_service_name(config.service_name.clone())
|
||||
.with_attribute(KeyValue::new("service.version", env!("CARGO_PKG_VERSION")))
|
||||
.build(),
|
||||
)
|
||||
.build();
|
||||
|
||||
// Set as global tracer provider
|
||||
@@ -191,7 +208,9 @@ pub async fn init_tracing(config: TracingConfig) -> CommonResult<()> {
|
||||
/// }
|
||||
/// ```
|
||||
pub async fn shutdown_tracing() {
|
||||
global::shutdown_tracer_provider();
|
||||
// In opentelemetry 0.31, shutdown is on the provider itself via Drop.
|
||||
// The global provider is shut down when it goes out of scope.
|
||||
// Explicit flush is no longer needed via a global function.
|
||||
}
|
||||
|
||||
/// Extract trace context from gRPC metadata.
|
||||
|
||||
@@ -65,7 +65,10 @@ spec:
|
||||
operator: In
|
||||
values:
|
||||
- "true"
|
||||
runtimeClassName: nvidia
|
||||
# NOTE: runtimeClassName: nvidia removed — Scaleway Kapsule GPU pools use
|
||||
# the nvidia container runtime as the default containerd handler. The
|
||||
# RuntimeClass CRD is only created by the full NVIDIA GPU Operator, not
|
||||
# the device plugin alone, so requiring it prevents pod scheduling.
|
||||
tolerations:
|
||||
# Tolerate GPU taints so the exporter can run on GPU nodes
|
||||
- operator: Exists
|
||||
|
||||
Reference in New Issue
Block a user