From dfdc91539567ed027fe2e8d803a31070992da7c1 Mon Sep 17 00:00:00 2001 From: jgrusewski Date: Tue, 28 Apr 2026 22:39:03 +0200 Subject: [PATCH] fix(argo): switch sanitizer/nsys templates to training-data-pvc (full 27 months) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The test-data-pvc only had 3-4 months of MBP-10 (Q1 2024 only), causing walk-forward fold generation to fail with 'Need at least 10 months of data'. Production training uses training-data-pvc mounted at /data with subdirs futures-baseline / futures-baseline-mbp10 / futures-baseline-trades — that's the full 27-month dataset. Switching all three env vars (FOXHUNT_TEST_DATA, FOXHUNT_MBP10_DATA, FOXHUNT_TRADES_DATA) and the volume/mount to point at the production data PVC. multi_fold_convergence's 6/2/2 walk-forward (10 months min) now has ample data. --- infra/k8s/argo/nsys-test-template.yaml | 18 ++++++++++-------- infra/k8s/argo/sanitizer-test-template.yaml | 19 +++++++++++-------- 2 files changed, 21 insertions(+), 16 deletions(-) diff --git a/infra/k8s/argo/nsys-test-template.yaml b/infra/k8s/argo/nsys-test-template.yaml index 174845395..d0e3eaa6b 100644 --- a/infra/k8s/argo/nsys-test-template.yaml +++ b/infra/k8s/argo/nsys-test-template.yaml @@ -65,9 +65,11 @@ spec: - name: cargo-target persistentVolumeClaim: claimName: cargo-target-cuda-test - - name: test-data + # Use the same training-data-pvc as production training (full 27 months + # of OHLCV + MBP-10 + trades) instead of the smaller test-data-pvc. + - name: training-data persistentVolumeClaim: - claimName: test-data-pvc + claimName: training-data-pvc readOnly: true templates: @@ -98,15 +100,15 @@ spec: - name: CUBLAS_WORKSPACE_CONFIG value: ":4096:8" - name: FOXHUNT_TEST_DATA - value: /data/test-data/ohlcv + value: /data/futures-baseline - name: TEST_DATA_DIR - value: /data/test-data/ohlcv + value: /data/futures-baseline # MBP-10 + trades are mandatory per feedback_mbp10_mandatory.md # (OFI features are part of the production state vector). - name: FOXHUNT_MBP10_DATA - value: /data/test-data/mbp10 + value: /data/futures-baseline-mbp10 - name: FOXHUNT_TRADES_DATA - value: /data/test-data/trades + value: /data/futures-baseline-trades - name: RUST_LOG value: info - name: LD_LIBRARY_PATH @@ -132,8 +134,8 @@ spec: readOnly: true - name: cargo-target mountPath: /cargo-target - - name: test-data - mountPath: /data/test-data + - name: training-data + mountPath: /data readOnly: true resources: requests: diff --git a/infra/k8s/argo/sanitizer-test-template.yaml b/infra/k8s/argo/sanitizer-test-template.yaml index c18de164d..85ca912f2 100644 --- a/infra/k8s/argo/sanitizer-test-template.yaml +++ b/infra/k8s/argo/sanitizer-test-template.yaml @@ -62,9 +62,12 @@ spec: - name: cargo-target persistentVolumeClaim: claimName: cargo-target-cuda-test - - name: test-data + # Use the same training-data-pvc as production training (full 27 months + # of OHLCV + MBP-10 + trades) instead of the smaller test-data-pvc which + # only has 3-4 months of MBP-10 — walk-forward needs ≥10 months. + - name: training-data persistentVolumeClaim: - claimName: test-data-pvc + claimName: training-data-pvc readOnly: true templates: @@ -95,15 +98,15 @@ spec: - name: CUBLAS_WORKSPACE_CONFIG value: ":4096:8" - name: FOXHUNT_TEST_DATA - value: /data/test-data/ohlcv + value: /data/futures-baseline - name: TEST_DATA_DIR - value: /data/test-data/ohlcv + value: /data/futures-baseline # MBP-10 + trades are mandatory per feedback_mbp10_mandatory.md # (OFI features are part of the production state vector). - name: FOXHUNT_MBP10_DATA - value: /data/test-data/mbp10 + value: /data/futures-baseline-mbp10 - name: FOXHUNT_TRADES_DATA - value: /data/test-data/trades + value: /data/futures-baseline-trades - name: RUST_LOG value: info - name: LD_LIBRARY_PATH @@ -114,8 +117,8 @@ spec: readOnly: true - name: cargo-target mountPath: /cargo-target - - name: test-data - mountPath: /data/test-data + - name: training-data + mountPath: /data readOnly: true resources: requests: