# DeployProof — bench ledger summary

Generated by `scripts/bench_extract.py summarize`. Raw rates only — attribution and narrative live in the report and dossiers.

## java-aws-reliability — 11/21 (52.4%) — ⚠ in-flight snapshot

> IN-FLIGHT snapshot — campaign live at extraction. The discovery protocol purges and re-runs rows whose failure class is platform-fixed (boundary-restarter), so listed failures include already-fixed classes pending re-run. Not citable as final.

| scenario | passed | n | rate |
|---|---|---|---|
| 01_spring_native_fargate | 2 | 2 | 100.0% |
| 02_spring_lambda_snapstart | 0 | 2 | 0.0% |
| 03_spring_jvm_fargate_ha | 3 | 3 | 100.0% |
| 04_quarkus_jvm_eks | 0 | 2 | 0.0% |
| 10_det_3vm_dev | 3 | 3 | 100.0% |
| 11_interactive_prod_generative | 1 | 2 | 50.0% |
| 12_learning_curve_java | 2 | 3 | 66.7% |
| 13_oos_prod_asg_alb | 0 | 2 | 0.0% |
| 14_spring_eks_generative | 0 | 2 | 0.0% |

## java-gcp-coverage — 11/12 (91.7%)

> 4 never-seen production topologies ×3, first contact. 11/12 gate, 12/12 deploy-health; scenario 21 recovered 0/3→3/3 via the headless-worker servlet fixer (2c2a0c78).

| scenario | passed | n | rate |
|---|---|---|---|
| 20_springboot_gke_hardened | 3 | 3 | 100.0% |
| 21_springboot_cloudrun_multiservice | 3 | 3 | 100.0% |
| 22_springboot_mig_https_lb | 2 | 3 | 66.7% |
| 23_springboot_gke_multizone_ha | 3 | 3 | 100.0% |

## java-gcp-reliability — 41/64 (64.1%)

> Raw discovery record, gate v1. 23 failures: 21 ENVIRONMENTAL / 1 HARNESS / 1 AGENT (Cloud Run probe-contract defect exposed by fix-then-rerun; fixed 92c5518d). Converged record: java-gcp-retest23.

| scenario | passed | n | rate |
|---|---|---|---|
| 01_java_cloudrun | 7 | 8 | 87.5% |
| 04_quarkus_jvm_gke | 6 | 8 | 75.0% |
| 10_det_3vm_dev | 7 | 8 | 87.5% |
| 11_interactive_prod_generative | 5 | 8 | 62.5% |
| 12_learning_curve_gcp | 3 | 8 | 37.5% |
| 12b_clean_3vm_gcp | 3 | 8 | 37.5% |
| 14_oos_prod_mig_global_lb | 5 | 8 | 62.5% |
| 98_learning_curve | 5 | 8 | 62.5% |

## java-gcp-retest23 — 23/23 (100.0%)

> Fix-then-rerun of Campaign A's 23 failed slots, final state. At first closure 22/23 (95.7%) with 01_java_cloudrun open; the class was root-caused (Google front end reserves /healthz externally) and platform-fixed (92c5518d); the 01 row is green on deploy-health basis (scale-to-zero idle window means recency-scoped logs are N/A, disclosed).

| scenario | passed | n | rate |
|---|---|---|---|
| 01_java_cloudrun | 1 | 1 | 100.0% |
| 04_quarkus_jvm_gke | 2 | 2 | 100.0% |
| 10_det_3vm_dev | 1 | 1 | 100.0% |
| 11_interactive_prod_generative | 3 | 3 | 100.0% |
| 12_learning_curve_gcp | 5 | 5 | 100.0% |
| 12b_clean_3vm_gcp | 5 | 5 | 100.0% |
| 14_oos_prod_mig_global_lb | 3 | 3 | 100.0% |
| 98_learning_curve | 3 | 3 | 100.0% |

## python-aws-reliability — 62/62 (100.0%) — ⚠ in-flight snapshot

> IN-FLIGHT snapshot at 62/64 reliability runs (coverage-12 queued behind it). 0 deploy failures on the ledger. READ WITH: (a) GATE MODE differs per scenario — 01/04/10/11/14 (40 rows) ran the strict 4-part gate (health + logs + metrics gating, connection measured); 12/12b/98 (22 rows) ran the learning-curve evaluation gate (endpoint reachability only, logs/metrics not gated) as their scenario specs declare. (b) 12_learning_curve_aws + 12b_clean_3vm_aws (15 rows so far) are declared generative but ROUTED DETERMINISTIC — the classifier judged the request complete and silently dropped the out-of-scope ask (an extra data volume per VM) because no schema field exists for it; they pass but do not exercise the generative path (open cross-cloud classifier defect, also seen on GCP). (c) 11_interactive_prod_generative routed deterministic as designed (full-spec request onto a template). Real generative evidence on this ledger = 98_learning_curve (7 rows, one of which needed 2 pre-apply verify/repair rounds). (d) Two off-ledger incidents are disclosed in the report: a model-provider credit exhaustion (402) that halted the harness cleanly, and a 2.5-hour worker wedge from a local DNS drop — both slots re-run green, neither is a deploy failure. Not citable as final.

| scenario | passed | n | rate |
|---|---|---|---|
| 01_py_fargate | 8 | 8 | 100.0% |
| 04_py_eks | 8 | 8 | 100.0% |
| 10_det_3vm_dev | 8 | 8 | 100.0% |
| 11_interactive_prod_generative | 8 | 8 | 100.0% |
| 12_learning_curve_aws | 8 | 8 | 100.0% |
| 12b_clean_3vm_aws | 7 | 7 | 100.0% |
| 14_oos_prod_asg_alb | 8 | 8 | 100.0% |
| 98_learning_curve | 7 | 7 | 100.0% |