Both images now use identical cuda:12.4.1-cudnn-runtime base, so maintaining
two separate Dockerfiles and build jobs was wasteful. Single image contains
all 7 training binaries, halving registry storage and Kaniko build time.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
RL models (DQN/PPO) only need basic CUDA runtime (~2GB), while supervised
models need cuDNN (~4GB). Splitting saves ~2GB pull time per RL job.
Rename the L4 GPU pool from ci-compile to ci-rl to reflect its actual use.
Add DaemonSet image pre-puller to cache training images on GPU nodes.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Changed from non-existent scw-bssd-nfs to scw-bssd (RWO).
Training output is single-pod, doesn't need ReadWriteMany.
PVC deployed to foxhunt namespace (WaitForFirstConsumer).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
RL env simulation is single-threaded (~1 core actual). Lowering from
7000m to 2000m request allows DQN + PPO hyperopt to run simultaneously
on one L4 node.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Fix ServiceAccount: use serviceAccount.name (not deprecated serviceAccountName)
- concurrent=1: one RL job gets full L4 node (7000m/7800m CPU, 16Gi/40Gi RAM)
- Faster iteration: ~2.3x more CPU per job vs splitting across 2 concurrent
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Set runtime_class_name=nvidia at runner level (all GPU pools have GPUs)
- Fix KUBERNETES_NODE_SELECTOR format: "key=value" not just value
- Replace wrong *_overwrite_allowed (regex) with correct
*_overwrite_max_allowed (max value) for CPU/memory overrides
- Remove per-job KUBERNETES_RUNTIME_CLASS_NAME (runner-level handles it)
- Simplify GPU detection in test/compile (GPU always present now)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
compile-services OOMKilled at 8Gi — ml crate with CUDA needs ~20Gi.
CI variable overrides (KUBERNETES_MEMORY_LIMIT) are silently dropped
by runner config merger for non-auto-generated fields. Increase
defaults to 3500m/12Gi request, 7800m/28Gi limit to fit L4 nodes.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Setting runtime_class_name = "" causes Kubernetes to reject pod creation
with "resource name may not be empty". Omitting it entirely means no
runtimeClassName on pod spec by default; GPU jobs override via
KUBERNETES_RUNTIME_CLASS_NAME CI variable.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
L4 ci-compile node was failing with 'no runtime for nvidia' because
fresh nodes take time to install nvidia drivers. Kaniko builds don't
need GPU at all. Now only Rust compile/test and training jobs request
nvidia RuntimeClass via KUBERNETES_RUNTIME_CLASS_NAME CI variable.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- compile-services now depends on check (not test), running in parallel
with test stage — saves ~10 min from critical path
- Runner concurrent bumped 4→10 so all 9 Kaniko builds run simultaneously
- Runner default resources lowered (500m/1Gi) for lightweight Kaniko jobs;
Rust compile/test/training override via KUBERNETES_*_REQUEST CI vars
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Previously nodeSelector was only applied via kubectl patch at deploy time.
Now it's in the YAML files to keep git and cluster in sync.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
All K8s nodeSelectors, CI config, runner config, monitoring docs, and
smoke tests now reference the new pool names provisioned by terragrunt.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
H100 is sm_90, not compatible with existing sm_89 sccache.
Revert runner node selector and resource limits to ci-build (L4).
Training validation stage (manual DQN/TFT smoke tests) stays.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Runner node selector: ci-build (L4) → gpu-training (H100) for faster
compilation (2x+ vCPUs) and native GPU training support
- Resource limits bumped: 20 vCPU / 128Gi (was 10 vCPU / 38Gi)
- New 'train' stage with manual DQN and TFT validation jobs
(50-step smoke tests to verify CUDA + model correctness)
- Requires helm upgrade of gitlab-runner with new values
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
CUDA cannot handle linear layers with zero input dimensions.
When num_static_features=0 or num_known_features=0, the
VariableSelectionNetwork constructor creates linear(0, 0, ...)
which triggers CUDA_ERROR_INVALID_VALUE on GPU.
- Make static/future VSN and GRN encoder fields Option<T>
- Skip layer creation when feature count is 0
- Forward pass skips absent feature paths gracefully
- Trainable adapter creates empty placeholder tensors
- Add cilium CNI toleration to training job template and runner
All 103 TFT tests pass.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Runner tags were only set during initial Helm registration but lost
on re-registration. Adding tag_list to the TOML config template
ensures tags survive runner pod restarts.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Switch output volume from PVC to emptyDir (sidecar uploads to S3)
- Map s3-credentials secret keys to AWS_ACCESS_KEY_ID/AWS_SECRET_ACCESS_KEY
(object_store crate expects standard AWS env var names)
- Fix data-dir path to include /futures-baseline subdirectory
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Adds a CronJob that runs on the 1st of every month at 03:00 UTC:
1. Reads current PAT from K8s Secret (gitlab-pat)
2. Calls GitLab rotate API (atomic: creates new, revokes old)
3. Updates K8s Secret with new token (3 retries)
4. Prints token to logs as recovery fallback if update fails
RBAC scoped to only get/patch the gitlab-pat secret.
Runs on gitlab node pool (zero GPU cost).
Manual trigger: kubectl -n foxhunt create job <name> --from=cronjob/gitlab-pat-rotation
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Extend ml_training_service to dispatch GPU training jobs as K8s batch/v1
Jobs, collect results via a Rust sidecar uploader, and support model
promotion with operator approval via fxt CLI.
- K8s dispatcher creates Jobs on gpu-training pool with native sidecar
- training_uploader crate: watches DONE/FAILED marker, uploads to S3,
reports completion via ReportJobCompletion gRPC
- PromotionManager compares metrics, queues better models for approval
- 4 new proto RPCs: ReportJobCompletion, ListPendingPromotions,
ApprovePromotion, RejectPromotion
- fxt commands: train start, model list/approve/reject
- Training binaries write DONE/FAILED markers + metrics.json
- Dockerfile, K8s job template, and CI pipeline updated
- StartTraining gracefully falls back to in-process when outside K8s
- 27 new tests (16 service + 11 promotion), 141 total service tests pass
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Adds a parent/child GitLab CI pipeline for ML model training:
- Generator script produces per-model hyperopt/train/evaluate jobs
- Parent pipeline (.gitlab-ci-training.yml) with manual trigger
- NFS-backed ReadWriteMany PVC for shared training outputs
- Hyperopt params wired into training binaries (DQN, PPO, TFT, Mamba2)
- Shared DBN loader eliminates duplicate code across hyperopt adapters
- Supervised hyperopt unified to DBN data (was parquet-only)
Pipeline: hyperopt (4 models) → train (10 models) → evaluate ensemble
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Delete 22 dead/placeholder/broken example files (-3,489 lines code)
- Delete 4 tracked CSV files (-1.1M lines, were accidentally committed)
- Move baseline training data default from data/cache/ to test_data/
- Update 5 unified binary defaults, gitignore, k8s upload comment, docs
- Consolidate all training data under test_data/futures-baseline/
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Create dedicated ci-build node pool (L4-1-24G, €0.75/hr) replacing
H100 (€3.50+/hr) for CI builds. H100 reserved for training only.
- Add ci-build pool to Terraform (L4-1-24G, scale-to-zero, fr-par-2)
- Update GitLab Runner to target ci-build pool with L4-sized limits
- Change CUDA_COMPUTE_CAP from 90 (H100) to 89 (L4) in CI
- Dockerfile.training keeps CUDA_COMPUTE_CAP=90 for H100 training
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add --cache=true --cache-repo to all 12 Kaniko builds
- Cache Docker layers in Scaleway CR (rg.fr-par.scw.cloud/foxhunt-ci/cache)
- Add Docker Hub auth to devcontainer + infra-runner prepare jobs
- First build populates cache; subsequent builds skip base image pulls
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Create 20Gi sccache-pvc (scw-bssd) for fast local cache hits
- Mount at /mnt/sccache in CI build pods via runner helm values
- Use SCCACHE_DIR instead of SCCACHE_BUCKET for check/test jobs
- Kaniko builds keep S3 sccache (can't mount PVCs in Kaniko)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Persistent block storage for Claude Code, cargo registry, sccache
cache, and shell config. Survives pod restarts and node scale-down.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Redis image WORKDIR is /data and entrypoint runs chown recursively.
PVC mounted at /data/training was read-only, causing chown to fail
with exit code 1, crashing the Redis sidecar.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Mount training-data-pvc at /data/training in build pods via runner config
- Update real_data_loader to search per-symbol subdirectories (Databento layout)
- Support .dbn.zst (zstd-compressed) files alongside raw .dbn
- Add FOXHUNT_DATA_DIR env var override for CI PVC path
- Extract decode_ohlcv_bars helper (generic over reader type)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Replace pod_spec strategic merge patch (silently not applied) with
runtime_class_name="nvidia". The nvidia RuntimeClass uses the
nvidia-container-runtime which injects GPU drivers, nvidia-smi, and
/dev/nvidia* devices into all containers automatically.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Strategic merge patches are applied to the Pod object, not PodSpec.
The patch needs {"spec":{"containers":[...]}} not {"containers":[...]}.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- poll_timeout 180s→600s: H100 provisioning takes ~3-5min from scale-to-zero
- build-ci-builder: only auto-run on push with Dockerfile changes (API
pipelines evaluate changes=true, causing unnecessary kaniko builds)
- check: add needs:[] to decouple from prepare stage
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
bindgen_cuda calls nvidia-smi to detect compute capability, which fails
without GPU device access. Setting CUDA_COMPUTE_CAP=90 bypasses this for
compilation. Pod spec GPU request ensures NVIDIA device plugin injects
/dev/nvidia* for test-time CUDA execution.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Remove orphaned `ci` pool (GP1-XS, 4 vCPU wasted — nothing scheduled to it).
Remove `ci-build` pool (GP1-M) — builds now run on gpu-training (H100-1-80G:
24 vCPU, 240GB, real CUDA). This eliminates the need for CUDA stubs,
separate test-gpu jobs, and ml crate exclusions.
Pool layout after:
always-on DEV1-M (core services, always on)
gitlab GP1-XS (GitLab CE + runner manager, always on)
gpu-training H100-1-80G (CI builds + ML training, scale-to-zero)
gpu-inference L4-1-24G (trading inference, scale-to-zero)
Build pod limits bumped to 16 vCPU / 64GB (from 6/12GB) to use H100 capacity.
Runner now has `gpu` tag — all tests including ml crate run in single job.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Add .gitignore entries for .dbn, .safetensors, .onnx and other large binaries
- Point GitLab Runner at internal cluster URL (not external Tailscale)
- Disable nginx proxy_buffering for git clone/push through Tailscale proxy
- Increase socat SSH proxy buffer sizes to 1MB
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Enable Prometheus server (GitLab chart bundled) on gitlab node pool
- Add socat SSH proxy sidecar (port 2222 → gitlab-shell) to Tailscale proxy
- Remove nginx stream module (not available in alpine) in favor of socat
- Set unlimited nginx client_max_body_size for large git pushes
- Add workhorse extraArgs for API limits
- Explicit always_on_type = DEV1-M in kapsule terragrunt
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Keep existing git.fxhnt.ai hostname for GitLab CE (repoint DNS
to new Tailscale IP after deploy). Add Terraform DNS module to
manage the A record via Terragrunt instead of manual scw CLI.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
