kubernetes-nmstate: add monitor tests to e2e-operator-ovn-ipv4

ci-operator/config/openshift/kubernetes-nmstate/openshift-kubernetes-nmstate-main.yaml

@@ -94,6 +94,12 @@ tests:
         IP_STACK=v4
         NETWORK_TYPE="OVNKubernetes"
         MIRROR_IMAGES=false
+    observers:
+      enable:
+      - observers-resource-watch
+    test:
+    - ref: kubernetes-nmstate-e2e-operator
+    - ref: kubernetes-nmstate-e2e-monitor-tests
     workflow: kubernetes-nmstate-e2e-operator
 - always_run: false
   as: e2e-handler-azure

ci-operator/step-registry/kubernetes-nmstate/e2e/monitor-tests/OWNERS

@@ -0,0 +1,11 @@
+approvers:
+- cybertron
+- phoracek
+- qinqon
+- rhrazdil
+reviewers:
+- cybertron
+- phoracek
+- qinqon
+- ramlavi
+- rhrazdil

ci-operator/step-registry/kubernetes-nmstate/e2e/monitor-tests/kubernetes-nmstate-e2e-monitor-tests-commands.sh

@@ -0,0 +1,43 @@
+#!/bin/bash
+
+set -euo pipefail
+
+export KUBECONFIG=${SHARED_DIR}/kubeconfig
+
+MONITOR_DIR="${ARTIFACT_DIR}/monitor-tests"
+mkdir -p "${MONITOR_DIR}"
+
+DISABLE_ARGS=""
+if [[ -n "${DISABLED_MONITOR_TESTS:-}" ]]; then
+  DISABLE_ARGS="--disable-monitor=${DISABLED_MONITOR_TESTS}"
+fi
+
+echo "Starting openshift-tests run-monitor..."
+openshift-tests run-monitor ${DISABLE_ARGS} --artifact-dir "${MONITOR_DIR}" &
+MONITOR_PID=$!
+
+echo "Monitoring cluster for ${MONITOR_DURATION} seconds..."
+sleep "${MONITOR_DURATION}"
+
+echo "Stopping monitor and evaluating results..."
+kill -TERM "${MONITOR_PID}" 2>/dev/null || true
+
+set +e
+wait "${MONITOR_PID}"
+monitor_exit=$?
+set -e
+
+echo "Monitor exited with code: ${monitor_exit}"
+
+# Copy junit results to artifact dir for CI reporting
+find "${MONITOR_DIR}" -name "junit*.xml" -exec cp {} "${ARTIFACT_DIR}/" \; 2>/dev/null || true
+
+# Exit code 143 (128+15) means SIGTERM was not caught by the process.
+# A well-behaved openshift-tests handles SIGTERM, evaluates results, and exits
+# with 0 (pass) or non-zero (failures detected).
+if [ "${monitor_exit}" -ne 0 ] && [ "${monitor_exit}" -ne 143 ]; then
+  echo "ERROR: Monitor tests detected failures (exit code: ${monitor_exit})"
+  exit 1
+fi
+
+echo "Monitor tests completed successfully."

ci-operator/step-registry/kubernetes-nmstate/e2e/monitor-tests/kubernetes-nmstate-e2e-monitor-tests-ref.metadata.json

@@ -0,0 +1,18 @@
+{
+	"path": "kubernetes-nmstate/e2e/monitor-tests/kubernetes-nmstate-e2e-monitor-tests-ref.yaml",
+	"owners": {
+		"approvers": [
+			"cybertron",
+			"phoracek",
+			"qinqon",
+			"rhrazdil"
+		],
+		"reviewers": [
+			"cybertron",
+			"phoracek",
+			"qinqon",
+			"ramlavi",
+			"rhrazdil"
+		]
+	}
+}

ci-operator/step-registry/kubernetes-nmstate/e2e/monitor-tests/kubernetes-nmstate-e2e-monitor-tests-ref.yaml

@@ -0,0 +1,23 @@
+ref:
+  as: kubernetes-nmstate-e2e-monitor-tests
+  from: tests
+  commands: kubernetes-nmstate-e2e-monitor-tests-commands.sh
+  env:
+  - name: MONITOR_DURATION
+    default: "60"
+    documentation: |-
+      Duration in seconds to run the monitor tests before evaluating results.
+  - name: DISABLED_MONITOR_TESTS
+    default: ""
+    documentation: |-
+      Comma-separated list of monitor tests to disable. Example:
+      "apiserver-new-disruption-invariant,disruption-summary-serializer"
+  resources:
+    requests:
+      cpu: 100m
+      memory: 200Mi
+  timeout: 30m0s
+  documentation: |-
+    Runs openshift-tests run-monitor for a configurable duration and fails the
+    job if any monitor tests detect issues such as pods using the default service
+    account, API server disruptions, or other cluster state violations.