Merge branch 'worktree-training-deploy' (CPU pipeline split)

This commit is contained in:
jgrusewski
2026-02-28 00:23:20 +01:00
2 changed files with 211 additions and 44 deletions

View File

@@ -1,8 +1,9 @@
# GitLab CI/CD — Foxhunt
# Two runners: CPU (compile/build/deploy) and GPU (training)
# CI builder image: CUDA 12.4 + Rust 1.89 + protoc + sccache (hosted on Scaleway CR)
# CI builder images:
# ci-builder: CUDA 12.4 + Rust 1.89 + protoc + sccache (for training binaries)
# ci-builder-cpu: Rust 1.89 + protoc + sccache, no CUDA (for service binaries)
# Service images: pushed to Scaleway Container Registry (Kaniko)
# Note: .gitlab-ci.yml itself is NOT in changes: filters — CI-only edits skip build
#
# Runner routing (by tags):
# kapsule,rust,docker → CPU runner → ci-compile-cpu pool (POP2-32C-128G, 32 vCPU)
@@ -50,6 +51,7 @@ variables:
AWS_SECRET_ACCESS_KEY: $SCW_SECRET_KEY
# CI builder image on Scaleway Container Registry (reachable by Kapsule nodes)
CI_BUILDER_IMAGE: rg.fr-par.scw.cloud/foxhunt-ci/ci-builder:latest
CI_BUILDER_CPU_IMAGE: rg.fr-par.scw.cloud/foxhunt-ci/ci-builder-cpu:latest
# Scaleway Container Registry for service images
REGISTRY: rg.fr-par.scw.cloud/foxhunt-ci
# IaC runner image on Scaleway Container Registry
@@ -87,6 +89,38 @@ build-ci-builder:
--cache=true --cache-repo="${REGISTRY}/cache"
--destination "${CI_BUILDER_IMAGE}"
# --------------------------------------------------------------------------
# Stage 0a: Build CPU-only CI builder image → push to Scaleway CR
# --------------------------------------------------------------------------
build-ci-builder-cpu:
stage: prepare
image:
name: gcr.io/kaniko-project/executor:debug
entrypoint: [""]
tags:
- kapsule
- docker
rules:
- if: $CI_PIPELINE_SOURCE == "push"
changes:
- infra/docker/Dockerfile.ci-builder-cpu
when: on_success
- when: manual
allow_failure: true
before_script:
- mkdir -p /kaniko/.docker
- >-
echo "{\"auths\":{
\"rg.fr-par.scw.cloud\":{\"username\":\"nologin\",\"password\":\"${SCW_SECRET_KEY}\"},
\"https://index.docker.io/v1/\":{\"username\":\"${DOCKERHUB_USERNAME}\",\"password\":\"${DOCKERHUB_TOKEN}\"}
}}" > /kaniko/.docker/config.json
script:
- /kaniko/executor
--context "${CI_PROJECT_DIR}"
--dockerfile "${CI_PROJECT_DIR}/infra/docker/Dockerfile.ci-builder-cpu"
--cache=true --cache-repo="${REGISTRY}/cache"
--destination "${CI_BUILDER_CPU_IMAGE}"
# --------------------------------------------------------------------------
# Stage 0b: Build devcontainer image → push to internal GitLab registry
# --------------------------------------------------------------------------
@@ -181,6 +215,29 @@ build-infra-runner:
echo "mold not found, falling back to lld via ld.mold wrapper"
fi
# Base template for CPU-only Rust jobs — lightweight builder without CUDA
# Uses ci-builder-cpu image (rust:1.89-slim, ~2-3GB, no CUDA toolkit)
.rust-base-cpu:
image: ${CI_BUILDER_CPU_IMAGE}
tags:
- kapsule
- rust
variables:
# No CUDA_COMPUTE_CAP — CPU-only builds don't compile CUDA kernels
CARGO_BUILD_JOBS: "30"
KUBERNETES_CPU_REQUEST: "28000m"
KUBERNETES_CPU_LIMIT: "31000m"
KUBERNETES_MEMORY_REQUEST: "24Gi"
KUBERNETES_MEMORY_LIMIT: "56Gi"
before_script:
# Mold linker fallback — same as .rust-base
- |
if ! command -v mold &>/dev/null; then
printf '#!/bin/sh\nexec ld.lld "$@"\n' > /usr/local/bin/ld.mold
chmod +x /usr/local/bin/ld.mold
echo "mold not found, falling back to lld via ld.mold wrapper"
fi
# --------------------------------------------------------------------------
# Stage 2: tests
# --------------------------------------------------------------------------
@@ -228,14 +285,13 @@ test:
- sccache --show-stats || true
# --------------------------------------------------------------------------
# Stage 3a: Compile all service binaries (single job, PVC sccache)
# Stage 3a: Compile service binaries (CPU-only, no CUDA)
# --------------------------------------------------------------------------
compile-services:
extends: .rust-base
extends: .rust-base-cpu
stage: compile
needs: []
rules:
# Only recompile when Rust source or Dockerfiles change — infra-only pushes skip
- if: $CI_COMMIT_BRANCH == "main" && $CI_PIPELINE_SOURCE == "push"
changes:
- Cargo.toml
@@ -246,16 +302,10 @@ compile-services:
- infra/docker/Dockerfile.*
- if: $CI_PIPELINE_SOURCE == "web" || $CI_PIPELINE_SOURCE == "api"
script:
# Compile uses CUDA stubs — no nvidia runtime needed, CUDA_COMPUTE_CAP from variables
- echo "CUDA_COMPUTE_CAP=$CUDA_COMPUTE_CAP"
- export SCCACHE_DIR="/mnt/sccache/sm_${CUDA_COMPUTE_CAP}"
- mkdir -p "$SCCACHE_DIR"
- sccache --zero-stats || true
# Select build profile: dev-release (fast, thin LTO) or release (full, fat LTO)
- mkdir -p "$SCCACHE_DIR" && sccache --zero-stats || true
- PROFILE=${DEV_RELEASE:+dev-release}; PROFILE=${PROFILE:-release}
- echo "Building with --profile $PROFILE"
- echo "Building services with --profile $PROFILE (CPU-only, no CUDA)"
- TARGET_DIR="target/$PROFILE"
# 1) Build service binaries (no CUDA feature)
- cargo build --profile $PROFILE
-p trading_service
-p api_gateway
@@ -265,7 +315,47 @@ compile-services:
-p trading_agent_service
-p data_acquisition_service
-p web-gateway
# 2) Build training binaries with CUDA (sccache already warm from step 1)
- sccache --show-stats || true
- mkdir -p build-out/services
- |
for bin in trading_service api_gateway broker_gateway_service ml_training_service backtesting_service trading_agent_service data_acquisition_service web-gateway; do
cp $TARGET_DIR/$bin build-out/services/
strip build-out/services/$bin
done
- ls -lh build-out/services/
artifacts:
paths:
- build-out/services/
expire_in: 1 hour
# --------------------------------------------------------------------------
# Stage 3a: Compile training binaries (CUDA stubs for candle GPU kernels)
# --------------------------------------------------------------------------
compile-training:
extends: .rust-base
stage: compile
needs: []
rules:
- if: $CI_COMMIT_BRANCH == "main" && $CI_PIPELINE_SOURCE == "push"
changes:
- Cargo.toml
- Cargo.lock
- crates/ml/**
- crates/trading_engine/**
- crates/common/**
- crates/risk/**
- crates/data/**
- crates/config/**
- crates/storage/**
- services/training_uploader/**
- if: $CI_PIPELINE_SOURCE == "web" || $CI_PIPELINE_SOURCE == "api"
script:
- echo "CUDA_COMPUTE_CAP=$CUDA_COMPUTE_CAP"
- export SCCACHE_DIR="/mnt/sccache/sm_${CUDA_COMPUTE_CAP}"
- mkdir -p "$SCCACHE_DIR" && sccache --zero-stats || true
- PROFILE=${DEV_RELEASE:+dev-release}; PROFILE=${PROFILE:-release}
- echo "Building training binaries with --profile $PROFILE (CUDA stubs)"
- TARGET_DIR="target/$PROFILE"
- cargo build --profile $PROFILE -p ml --features ml/cuda
--example train_baseline_rl
--example train_baseline_supervised
@@ -273,31 +363,28 @@ compile-services:
--example evaluate_supervised
--example hyperopt_baseline_rl
--example hyperopt_baseline_supervised
# 3) Build training uploader sidecar (no CUDA needed)
- cargo build --profile $PROFILE -p training_uploader
- sccache --show-stats || true
- mkdir -p build-out
- mkdir -p build-out/training
- |
for bin in trading_service api_gateway broker_gateway_service ml_training_service backtesting_service trading_agent_service data_acquisition_service web-gateway; do
cp $TARGET_DIR/$bin build-out/
strip build-out/$bin
done
for bin in train_baseline_rl train_baseline_supervised evaluate_baseline evaluate_supervised hyperopt_baseline_rl hyperopt_baseline_supervised; do
cp $TARGET_DIR/examples/$bin build-out/
strip build-out/$bin
cp $TARGET_DIR/examples/$bin build-out/training/
strip build-out/training/$bin
done
cp $TARGET_DIR/training_uploader build-out/
strip build-out/training_uploader
- ls -lh build-out/
cp $TARGET_DIR/training_uploader build-out/training/
strip build-out/training/training_uploader
- ls -lh build-out/training/
artifacts:
paths:
- build-out/
- build-out/training/
expire_in: 1 hour
# --------------------------------------------------------------------------
# Stage 3b: Package service images (Kaniko → Scaleway CR, uses pre-built binaries)
# --------------------------------------------------------------------------
.kaniko-base:
# Kaniko base for service images (depends on compile-services)
.kaniko-service-base:
stage: build
needs:
- job: compile-services
@@ -332,72 +419,105 @@ compile-services:
\"https://index.docker.io/v1/\":{\"username\":\"${DOCKERHUB_USERNAME}\",\"password\":\"${DOCKERHUB_TOKEN}\"}
}}" > /kaniko/.docker/config.json
# Kaniko base for training image (depends on compile-training)
.kaniko-training-base:
stage: build
needs:
- job: compile-training
image:
name: gcr.io/kaniko-project/executor:debug
entrypoint: [""]
tags:
- kapsule
- docker
rules:
- if: $CI_COMMIT_BRANCH == "main" && $CI_PIPELINE_SOURCE == "push"
changes:
- Cargo.toml
- Cargo.lock
- crates/ml/**
- crates/trading_engine/**
- crates/common/**
- crates/risk/**
- crates/data/**
- crates/config/**
- crates/storage/**
- services/training_uploader/**
- if: $CI_PIPELINE_SOURCE == "web" || $CI_PIPELINE_SOURCE == "api"
before_script:
- mkdir -p /kaniko/.docker
- >-
echo "{\"auths\":{
\"rg.fr-par.scw.cloud\":{\"username\":\"nologin\",\"password\":\"${SCW_SECRET_KEY}\"},
\"https://index.docker.io/v1/\":{\"username\":\"${DOCKERHUB_USERNAME}\",\"password\":\"${DOCKERHUB_TOKEN}\"}
}}" > /kaniko/.docker/config.json
# 7 services: pre-built binary → Dockerfile.runtime (no Rust toolchain, ~30s each)
build-trading-service:
extends: .kaniko-base
extends: .kaniko-service-base
script:
- /kaniko/executor
--context dir://${CI_PROJECT_DIR}/build-out
--context dir://${CI_PROJECT_DIR}/build-out/services
--dockerfile "${CI_PROJECT_DIR}/infra/docker/Dockerfile.runtime"
--build-arg SERVICE=trading_service
--destination "${REGISTRY}/trading_service:${CI_COMMIT_SHA}"
--destination "${REGISTRY}/trading_service:latest"
build-api-gateway:
extends: .kaniko-base
extends: .kaniko-service-base
script:
- /kaniko/executor
--context dir://${CI_PROJECT_DIR}/build-out
--context dir://${CI_PROJECT_DIR}/build-out/services
--dockerfile "${CI_PROJECT_DIR}/infra/docker/Dockerfile.runtime"
--build-arg SERVICE=api_gateway
--destination "${REGISTRY}/api_gateway:${CI_COMMIT_SHA}"
--destination "${REGISTRY}/api_gateway:latest"
build-broker-gateway:
extends: .kaniko-base
extends: .kaniko-service-base
script:
- /kaniko/executor
--context dir://${CI_PROJECT_DIR}/build-out
--context dir://${CI_PROJECT_DIR}/build-out/services
--dockerfile "${CI_PROJECT_DIR}/infra/docker/Dockerfile.runtime"
--build-arg SERVICE=broker_gateway_service
--destination "${REGISTRY}/broker_gateway_service:${CI_COMMIT_SHA}"
--destination "${REGISTRY}/broker_gateway_service:latest"
build-ml-training:
extends: .kaniko-base
extends: .kaniko-service-base
script:
- /kaniko/executor
--context dir://${CI_PROJECT_DIR}/build-out
--context dir://${CI_PROJECT_DIR}/build-out/services
--dockerfile "${CI_PROJECT_DIR}/infra/docker/Dockerfile.runtime"
--build-arg SERVICE=ml_training_service
--destination "${REGISTRY}/ml_training_service:${CI_COMMIT_SHA}"
--destination "${REGISTRY}/ml_training_service:latest"
build-backtesting:
extends: .kaniko-base
extends: .kaniko-service-base
script:
- /kaniko/executor
--context dir://${CI_PROJECT_DIR}/build-out
--context dir://${CI_PROJECT_DIR}/build-out/services
--dockerfile "${CI_PROJECT_DIR}/infra/docker/Dockerfile.runtime"
--build-arg SERVICE=backtesting_service
--destination "${REGISTRY}/backtesting_service:${CI_COMMIT_SHA}"
--destination "${REGISTRY}/backtesting_service:latest"
build-trading-agent:
extends: .kaniko-base
extends: .kaniko-service-base
script:
- /kaniko/executor
--context dir://${CI_PROJECT_DIR}/build-out
--context dir://${CI_PROJECT_DIR}/build-out/services
--dockerfile "${CI_PROJECT_DIR}/infra/docker/Dockerfile.runtime"
--build-arg SERVICE=trading_agent_service
--destination "${REGISTRY}/trading_agent_service:${CI_COMMIT_SHA}"
--destination "${REGISTRY}/trading_agent_service:latest"
build-data-acquisition:
extends: .kaniko-base
extends: .kaniko-service-base
script:
- /kaniko/executor
--context dir://${CI_PROJECT_DIR}/build-out
--context dir://${CI_PROJECT_DIR}/build-out/services
--dockerfile "${CI_PROJECT_DIR}/infra/docker/Dockerfile.runtime"
--build-arg SERVICE=data_acquisition_service
--destination "${REGISTRY}/data_acquisition_service:${CI_COMMIT_SHA}"
@@ -405,7 +525,7 @@ build-data-acquisition:
# web-gateway: Node.js dashboard build + pre-built Rust binary
build-web-gateway:
extends: .kaniko-base
extends: .kaniko-service-base
script:
- /kaniko/executor
--context "${CI_PROJECT_DIR}"
@@ -416,10 +536,10 @@ build-web-gateway:
# training: pre-built CUDA binaries → CUDA runtime image (no recompilation)
build-training:
extends: .kaniko-base
extends: .kaniko-training-base
script:
- /kaniko/executor
--context dir://${CI_PROJECT_DIR}/build-out
--context dir://${CI_PROJECT_DIR}/build-out/training
--dockerfile "${CI_PROJECT_DIR}/infra/docker/Dockerfile.training-runtime"
--destination "${REGISTRY}/training:${CI_COMMIT_SHA}"
--destination "${REGISTRY}/training:latest"

View File

@@ -0,0 +1,47 @@
# CUDA-free CI builder image for Foxhunt service compilation
# ~2-3GB vs ~8GB for the CUDA devel variant (Dockerfile.ci-builder)
# Contains: Rust 1.89, protoc, sccache, git, OpenSSL, clang, mold, lld, make
# Build: docker build -f infra/docker/Dockerfile.ci-builder-cpu -t foxhunt-ci-builder-cpu .
# Push: docker tag foxhunt-ci-builder-cpu rg.fr-par.scw.cloud/foxhunt-ci/ci-builder-cpu:latest
# docker push rg.fr-par.scw.cloud/foxhunt-ci/ci-builder-cpu:latest
FROM rust:1.89-slim-bookworm
ENV DEBIAN_FRONTEND=noninteractive
RUN apt-get update && apt-get install -y --no-install-recommends \
git \
curl \
ca-certificates \
pkg-config \
libssl-dev \
lld \
clang \
make \
perl \
build-essential \
libfontconfig1-dev \
unzip \
&& rm -rf /var/lib/apt/lists/*
# mold linker — 2-5x faster than lld for final link step on large Rust binaries
RUN curl -fsSL https://github.com/rui314/mold/releases/download/v2.35.1/mold-2.35.1-x86_64-linux.tar.gz \
| tar xz -C /usr/local --strip-components=1 \
&& chmod +x /usr/local/bin/mold
# protoc 28.3 (proto3 optional support, matches local dev)
RUN curl -fsSL https://github.com/protocolbuffers/protobuf/releases/download/v28.3/protoc-28.3-linux-x86_64.zip -o /tmp/protoc.zip \
&& unzip -o /tmp/protoc.zip -d /usr/local bin/protoc 'include/*' \
&& rm /tmp/protoc.zip \
&& chmod +x /usr/local/bin/protoc
# sccache for S3-backed build caching
RUN curl -fsSL https://github.com/mozilla/sccache/releases/download/v0.10.0/sccache-v0.10.0-x86_64-unknown-linux-musl.tar.gz \
| tar xz -C /usr/local/bin --strip-components=1 sccache-v0.10.0-x86_64-unknown-linux-musl/sccache \
&& chmod +x /usr/local/bin/sccache
# clippy pre-installed
RUN rustup component add clippy
# Verify
RUN rustc --version && cargo --version && git --version && protoc --version && sccache --version && make --version && mold --version