[CI] Per-model, per-entry-mode code coverage collection (#5593)

Signed-off-by: Shenglei Fu <sfu@confluent.io>
Signed-off-by: Shenglei Fu <117230642+ShengleiFu@users.noreply.github.com>
Co-authored-by: Claude Sonnet 5 <noreply@anthropic.com>
Co-authored-by: Alicia <115451386+congw729@users.noreply.github.com>
Co-authored-by: wangyu <53896905+yenuo26@users.noreply.github.com>
Co-authored-by: Hongsheng Liu <liuhongsheng4@huawei.com>

.buildkite/common/ci_mirror_hardwares.yml

@@ -9,6 +9,11 @@
 #   a2b3_npu_1  — ascend-a2b3, resource_class npu-1 (A2/B3 CI image)
 #   a2b3_npu_4  — ascend-a2b3, resource_class npu-4
 #   a3_npu_2    — ascend-a3, resource_class npu-2
+#
+# The ``l4_*`` presets run under the ``docker`` plugin, which does not put
+# ``buildkite-agent`` on PATH inside the container the way the ``kubernetes``
+# presets do, so they set ``mount-buildkite-agent: true``; without it any step
+# calling ``buildkite-agent`` (artifact upload, annotate, meta-data) fails there.
 
 x-npu-k8s-plugins: &npu_k8s_plugins
   - kubernetes:
@@ -26,6 +31,7 @@ mirror_hardwares:
           image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
           always-pull: true
           propagate-environment: true
+          mount-buildkite-agent: true
           shm-size: "8gb"
           environment:
             - HF_HOME=/fsx/hf_cache
@@ -41,6 +47,7 @@ mirror_hardwares:
           image: public.ecr.aws/q9t5s3a7/vllm-ci-test-repo:$BUILDKITE_COMMIT
           always-pull: true
           propagate-environment: true
+          mount-buildkite-agent: true
           shm-size: "8gb"
           environment:
             - HF_HOME=/fsx/hf_cache

.buildkite/common/scripts/run_cov_split.sh

@@ -0,0 +1,82 @@
+#!/usr/bin/env bash
+# Run a model's E2E tests once per entry mode as separate coverage runs and upload
+# the reports, so per-mode attribution survives (see vllm-project/vllm-omni#5332).
+#
+# Usage: run_cov_split.sh --model-id <id> [--offline <paths>] [--online <paths>] \
+#                         -- <pytest args...>
+#
+#   --model-id  Names the artifacts: coverage-<id>-{offline,online}-<step id>.xml.gz.
+#               Use the model's directory name under vllm_omni/*/models/.
+#   --offline   Pytest path(s) for the offline half. Repeatable.
+#   --online    Pytest path(s) for the online half. Repeatable.
+#   --          Everything after this is forwarded verbatim to every pytest run,
+#               quoting preserved: the job's own -m expression, --run-level, and
+#               anything else it needs (e.g. --test-config-file).
+#
+# At least one of --offline/--online is required, so a job with only one mode runs
+# just that half. Total runtime is bounded by the step's timeout_in_minutes.
+#
+# Exits non-zero if any half or the upload failed.
+set -uo pipefail
+
+MODEL_ID=""
+OFFLINE=()
+ONLINE=()
+PYTEST_ARGS=()
+
+while [[ $# -gt 0 ]]; do
+    case "$1" in
+        --model-id) MODEL_ID="$2"; shift 2 ;;
+        --offline)  OFFLINE+=("$2"); shift 2 ;;
+        --online)   ONLINE+=("$2"); shift 2 ;;
+        --)         shift; PYTEST_ARGS=("$@"); break ;;
+        *) echo "run_cov_split.sh: unknown argument: $1" >&2; exit 2 ;;
+    esac
+done
+
+if [[ -z "${MODEL_ID}" ]]; then
+    echo "run_cov_split.sh: missing --model-id" >&2
+    exit 2
+fi
+
+if ((${#OFFLINE[@]} == 0 && ${#ONLINE[@]} == 0)); then
+    echo "run_cov_split.sh: need at least one of --offline / --online" >&2
+    exit 2
+fi
+
+# Two steps can test the same model, and artifacts are scoped to the build (a
+# nightly one carries the ready, merge and nightly tiers at once), so unsuffixed
+# reports would share a path -- a glob download then races to rename onto one file.
+# Not BUILDKITE_JOB_ID: the step id survives retries, so attempts collapse under
+# Buildkite's latest-attempt lookup instead of accumulating. A parallel step would
+# need BUILDKITE_PARALLEL_JOB as well; no pilot is one.
+REPORT_SUFFIX="${BUILDKITE_STEP_ID:-local}"
+OFFLINE_REPORT="coverage-${MODEL_ID}-offline-${REPORT_SUFFIX}.xml"
+ONLINE_REPORT="coverage-${MODEL_ID}-online-${REPORT_SUFFIX}.xml"
+
+run_half() {
+    local mode="$1"
+    shift
+    local report="coverage-${MODEL_ID}-${mode}-${REPORT_SUFFIX}.xml"
+    echo "--- coverage: ${MODEL_ID} ${mode}"
+    pytest -s -v "$@" "${PYTEST_ARGS[@]}" --cov=vllm_omni --cov-report="xml:${report}"
+    local status=$?
+    # Cobertura lists every statement in vllm_omni whether or not it ran, so the
+    # report is ~7MB no matter how little the job covers. The repetition is what
+    # dominates it, and gzip takes that to ~400KB.
+    gzip -9 -f "${report}" || status=1
+    return "${status}"
+}
+
+# Clear both modes, not just the ones being run: the upload glob matches both, so
+# a report left by an earlier run would otherwise be published as if the mode it
+# belongs to had run this time.
+rm -f "${OFFLINE_REPORT}" "${ONLINE_REPORT}" "${OFFLINE_REPORT}.gz" "${ONLINE_REPORT}.gz"
+
+EXIT=0
+((${#OFFLINE[@]})) && { run_half offline "${OFFLINE[@]}" || EXIT=1; }
+((${#ONLINE[@]})) && { run_half online "${ONLINE[@]}" || EXIT=1; }
+
+buildkite-agent artifact upload "coverage-${MODEL_ID}-*-${REPORT_SUFFIX}.xml.gz" || EXIT=1
+
+exit "${EXIT}"

.buildkite/cuda/test-merge.yml

@@ -94,13 +94,19 @@ steps:
           - vllm_omni/model_executor/models/qwen3_tts/
           - vllm_omni/model_executor/stage_input_processors/qwen3_tts.py
           - vllm_omni/deploy/qwen3_tts.yaml
+          - .buildkite/common/scripts/run_cov_split.sh
+          - pyproject.toml
+        # 40 -> 60: the coverage split runs the model twice.
+        timeout_in_minutes: 60
         commands:
           - |
-            timeout 40m bash -c "
-              export VLLM_LOGGING_LEVEL=DEBUG
-              export VLLM_ALLOW_LONG_MAX_MODEL_LEN=1
-              pytest -s -v tests/e2e/online_serving/test_qwen3_tts_base.py tests/e2e/offline_inference/test_qwen3_tts_base.py -m 'advanced_model and cuda' --run-level 'advanced_model'
-            "
+            export VLLM_LOGGING_LEVEL=DEBUG
+            export VLLM_ALLOW_LONG_MAX_MODEL_LEN=1
+            .buildkite/common/scripts/run_cov_split.sh \
+              --model-id qwen3_tts \
+              --offline tests/e2e/offline_inference/test_qwen3_tts_base.py \
+              --online tests/e2e/online_serving/test_qwen3_tts_base.py \
+              -- -m 'advanced_model and cuda' --run-level 'advanced_model'
         mirror_hardwares: l4_1
 
       - label: "TTS · Higgs-Audio-V3 Test"
@@ -203,9 +209,17 @@ steps:
           - vllm_omni/diffusion/models/bagel/
           - vllm_omni/model_executor/stage_input_processors/bagel.py
           - vllm_omni/deploy/bagel.yaml
-        timeout_in_minutes: 20
+          - .buildkite/common/scripts/run_cov_split.sh
+          - pyproject.toml
+        # 20 -> 40: the coverage split runs the model twice.
+        timeout_in_minutes: 40
         commands:
-          - pytest -s -v tests/e2e/offline_inference/test_bagel.py tests/e2e/online_serving/test_bagel.py -m 'advanced_model and cuda' --run-level 'advanced_model'
+          - |
+            .buildkite/common/scripts/run_cov_split.sh \
+              --model-id bagel \
+              --offline tests/e2e/offline_inference/test_bagel.py \
+              --online tests/e2e/online_serving/test_bagel.py \
+              -- -m 'advanced_model and cuda' --run-level 'advanced_model'
         mirror_hardwares: h100_1
 
       - label: "Diffusion · Qwen Image Test"

.gitignore

@@ -46,6 +46,8 @@ htmlcov/
 .cache
 nosetests.xml
 coverage.xml
+coverage-*.xml
+coverage-*.xml.gz
 *.cover
 *.py,cover
 .hypothesis/

docs/contributing/ci/ci_settings.md

@@ -327,6 +327,96 @@ On PR builds, `upload_pipeline.py` logs `skip '…' (no changes under …)` for
 - Use **groups** for dashboard readability; keep **E2E Test** as the group name CUDA diff filtering expects for `--e2e` nightly runs on main.
 - Prefix labels by model domain: **Omni ·**, **TTS ·**, **Diffusion ·**, **Simple ·**, etc., matching existing steps.
 
+### Per-model coverage
+
+Pilot (v1): two Merge-tier (L3) jobs in `.buildkite/cuda/test-merge.yml` upload
+per-model, per-entry-mode coverage as Buildkite artifacts — "Diffusion · Bagel
+Test" and "TTS · Qwen3-TTS Base Test". Both run on a single GPU so the pilot is
+cheap to reproduce. This is a pilot, not a rollout: no dashboard/visualization
+lives in this repo, and nothing is gated on coverage.
+
+**Naming convention**: `coverage-<model_id>-<mode>-<step_id>.xml.gz`, where
+`<model_id>` is the model's directory name under
+`vllm_omni/diffusion/models/<model_id>/` or
+`vllm_omni/model_executor/models/<model_id>/` (e.g. `z_image`, `qwen3_omni`),
+`<mode>` is `online` or `offline`, and `<step_id>` is `BUILDKITE_STEP_ID`
+(`local` outside Buildkite). `gunzip` before feeding a report to a Cobertura
+consumer.
+
+Reports are gzipped because `--cov=vllm_omni` sets coverage's `source`, which
+makes it walk the package and emit one `<line>` element per statement for every
+file — including the ones a single model's tests never import. That inventory,
+not the coverage, is the size: a Bagel report is 7.1 MB, of which 96.5% is
+`<line>` elements and 77% belongs to the 840 of 1169 files that run never
+imported. A run covering zero lines produces the same 7 MB. gzip takes it to
+roughly 410 KB and changes nothing about the data.
+
+`<step_id>` is what keeps two steps that cover the same model apart. Artifacts are
+scoped to the build, and a nightly build carries the ready, merge and nightly tiers
+at once, so several same-model steps land in one namespace — `qwen3_tts` already has
+both a Base and a CustomVoice job, and `minicpmo_4_5` a base and a duplex one. Two
+uploads on one path do not fail loudly: an exact-name download can report the path
+as ambiguous, and a glob download fetches every match concurrently and renames each
+onto the same destination, so whichever finishes last wins. The step id remains the
+logical step identifier across retries, while Buildkite's default artifact lookup
+selects only the latest attempt — `BUILDKITE_JOB_ID` would instead give every
+attempt its own filename, leaving a failed attempt's partial report to be picked up
+alongside the real one. A step using `parallelism`/matrix would need
+`BUILDKITE_PARALLEL_JOB` added, since its jobs share a step id; neither pilot is
+one.
+
+**Opting in a new model**: replace the job's combined `pytest` command with
+[`run_cov_split.sh`](https://github.com/vllm-project/vllm-omni/blob/main/.buildkite/common/scripts/run_cov_split.sh),
+which runs each mode, names the reports, uploads them, and fails the step if
+either half or the upload failed:
+
+```yaml
+commands:
+  - |
+    .buildkite/common/scripts/run_cov_split.sh \
+      --model-id <model_id> \
+      --offline <offline test path> \
+      --online <online test path> \
+      -- <the job's existing pytest flags, e.g. -m '...' --run-level '...'>
+```
+
+Everything after `--` is forwarded verbatim to both runs with quoting preserved,
+so an expression like `-m 'advanced_model and cuda'` survives intact and a job can
+pass anything else it needs. `--offline`/`--online` are repeatable and only one is
+required, so a job with tests in a single mode runs just that half. Runtime is
+bounded solely by the step's `timeout_in_minutes` — raise it, since the split
+loads the model once per mode.
+
+Before splitting, confirm each half actually collects at least one test under the job's forwarded filters — pytest exits with code 5 ("no tests collected") if a half is empty, which the script reports as a failure, red-ing a job that used to pass as one combined invocation.
+
+Also check the tests' `num_cards` against the job's `mirror_hardwares` preset.
+`cuda_marks` turns `num_cards > 1` into `skipif(device_count() < num_cards)`, so a
+test asking for more GPUs than the preset provides is silently skipped and its
+coverage artifact comes back empty. Splitting such a job produces a file that
+looks fine and measures nothing — "Diffusion · Z Image Test" is the current
+example (online tests want 4 cards, the job runs on `l4_1`), which is why it is
+not a coverage pilot.
+
+Online-mode coverage depends on the `[tool.coverage.run]` `patch`/`sigterm`
+settings in `pyproject.toml`: online tests launch the server with
+`subprocess.Popen` and stop it with SIGTERM, and without those settings the XML
+reflects only the pytest parent process, not the server's code paths.
+
+The upload depends on `buildkite-agent` being callable inside the container. The
+`kubernetes` presets (`h100_*`, `*_npu_*`) provide it; the `docker` ones (`l4_*`)
+only do because they set `mount-buildkite-agent: true`. A new docker preset that
+runs a coverage job needs the same.
+
+List both `run_cov_split.sh` and `pyproject.toml` in every opted-in job's
+`source_file_dependencies` — both change what the job measures, so without them a
+change there is filtered out of normal PR builds and only surfaces in a later
+nightly. `tests/buildkite/test_upload_pipeline.py` asserts this for the pilots.
+Editing only the surrounding CI YAML still does not schedule them, so a PR that
+touches just the wiring needs a full E2E run (or the commands run on a GPU host)
+to produce artifacts. When checking a new model's
+artifacts, compare `lines-covered` between the online and offline XML rather than
+just confirming both files exist.
+
 ### Validation checklist
 
 | Check | Command / location |

pyproject.toml

@@ -301,6 +301,21 @@ filterwarnings = [
     "ignore:.*does not have '__bases__' attribute.*:UserWarning",
 ]
 
+[tool.coverage.run]
+core = "sysmon"
+source = ["vllm_omni"]
+# Online-serving tests launch the server via subprocess.Popen and stop it with
+# SIGTERM. pytest-cov no longer instruments subprocesses implicitly, and a
+# SIGTERM-ed process discards its data, so without patch/sigterm a coverage run
+# over an online test measures only the pytest parent.
+patch = ["subprocess"]
+concurrency = ["multiprocessing", "thread"]
+sigterm = true
+
+[tool.coverage.report]
+show_missing = true
+skip_covered = true
+
 [tool.typos.default]
 extend-ignore-identifiers-re = [
     ".*_thw",

tests/buildkite/test_upload_pipeline.py

@@ -5,6 +5,7 @@ import sys
 from pathlib import Path
 
 import pytest
+import yaml
 
 sys.path.insert(0, str(Path(__file__).resolve().parents[2] / ".buildkite" / "common" / "scripts"))
 
@@ -139,3 +140,48 @@ def test_mirror_hardwares_a2b3_npu_4_expands_agents_image_and_plugins() -> None:
     assert step["plugins"][0]["kubernetes"]["podSpecPatch"]["imagePullSecrets"] == [
         {"name": "swr-secret"},
     ]
+
+
+COVERAGE_PILOT_LABELS = ("Diffusion · Bagel Test", "TTS · Qwen3-TTS Base Test")
+MERGE_PIPELINE = Path(".buildkite/cuda/test-merge.yml")
+
+
+def _surviving_labels(changed_files: list[str]) -> set[str]:
+    doc = yaml.safe_load(MERGE_PIPELINE.read_text(encoding="utf-8"))
+    rendered = _render_test_pipeline(doc, changed_files=changed_files)
+    labels: set[str] = set()
+
+    def walk(steps: list | None) -> None:
+        for step in steps or []:
+            if not isinstance(step, dict):
+                continue
+            if "label" in step:
+                labels.add(step["label"])
+            walk(step.get("steps"))
+
+    walk(rendered.get("steps"))
+    return labels
+
+
+@pytest.mark.parametrize(
+    "changed_file",
+    [
+        ".buildkite/common/scripts/run_cov_split.sh",
+        "pyproject.toml",
+    ],
+)
+def test_coverage_pilots_selected_by_their_shared_inputs(changed_file: str) -> None:
+    """The split helper and the coverage config both change what these jobs measure.
+
+    Without them in source_file_dependencies the jobs are filtered out of normal
+    PR builds, so a regression would only surface in a later nightly.
+    """
+    labels = _surviving_labels([changed_file])
+    for pilot in COVERAGE_PILOT_LABELS:
+        assert pilot in labels, f"{pilot} dropped when {changed_file} changed"
+
+
+def test_coverage_pilots_not_selected_by_unrelated_change() -> None:
+    labels = _surviving_labels(["vllm_omni/entrypoints/openai/serving_chat.py"])
+    for pilot in COVERAGE_PILOT_LABELS:
+        assert pilot not in labels