[env_op_images] Add CRI-O pull verification to pulled-images report

Cross-reference the pulled-images report with CRI-O journal logs
from cluster nodes to confirm which images were actually pulled by
the container runtime. Runs automatically when kubeconfig is defined,
same as the pulled-images report itself.

Co-authored-by: Cursor <cursor@cursor.com>

Signed-off-by: nemarjan <nemarjan@redhat.com>

Author: nemarjan

plugins/modules/verify_pulled_report_crio.py

@@ -0,0 +1,298 @@
+#!/usr/bin/python
+
+# Copyright: (c) 2026, Nemanja Marjanovic <nemarjan@redhat.com>
+# Apache License Version 2.0 (see LICENSE)
+
+from __future__ import absolute_import, division, print_function
+
+__metaclass__ = type
+
+DOCUMENTATION = r"""
+---
+module: verify_pulled_report_crio
+
+short_description: Enrich pulled_images_report with CRI-O pull evidence
+
+description:
+  - Reads the YAML produced by the env_op_images pulled-images report role task.
+  - "Parses CRI-O journal lines for C(msg=\"Pulled image: ...@sha256:...\")."
+  - Adds per-row verification fields using trusted mirror domains from
+    C(summary.mirror_rules).
+  - When images carry a C(node) field, evidence is matched against the
+    specific node's CRI-O log first.  If the digest is only found on a
+    different node the status is set to C(MISMATCH_CROSS_NODE).
+  - Log files are expected to follow the C(<node-name>.crio.log) naming
+    convention produced by the role task.
+
+options:
+  report_path:
+    description: Path to C(pulled_images_report.yaml) (input).
+    required: true
+    type: str
+  output_path:
+    description: Path for the enriched YAML report (output).
+    required: true
+    type: str
+  log_paths:
+    description:
+      - Explicit list of log files to parse (e.g. per-node CRI-O logs).
+      - Combined with files found under I(log_dir) when set.
+    required: false
+    type: list
+    elements: str
+    default: []
+  log_dir:
+    description:
+      - Directory containing CRI-O log files matching I(log_glob).
+    required: false
+    type: str
+  log_glob:
+    description: Glob under I(log_dir). Used only when I(log_dir) is set.
+    required: false
+    default: "*.crio.log"
+    type: str
+
+author:
+  - Nemanja Marjanovic (@nemarjan)
+
+notes:
+  - Requires PyYAML on the controller (same as other cifmw.general modules).
+"""
+
+EXAMPLES = r"""
+- name: Enrich pulled report using fetched node logs
+  cifmw.general.verify_pulled_report_crio:
+    report_path: "{{ cifmw_env_op_images_pulled_report_path }}"
+    log_dir: "{{ cifmw_env_op_images_crio_logs_dir }}"
+    output_path: "{{ cifmw_env_op_images_verified_report_path }}"
+"""
+
+RETURN = r"""
+changed:
+  description: Whether the output file was written.
+  type: bool
+  returned: always
+trusted_mirrors:
+  description: Hostnames extracted from mirror rules in the report summary.
+  type: list
+  elements: str
+  returned: always
+log_files:
+  description: Number of log files read.
+  type: int
+  returned: always
+entries_with_digest:
+  description: Image rows that had a sha256 digest in C(image_id).
+  type: int
+  returned: always
+cross_node_entries:
+  description: >-
+    Image rows where evidence was found only on a different node
+    than where the pod ran (C(MISMATCH_CROSS_NODE)).
+  type: int
+  returned: always
+nodes_with_evidence:
+  description: >-
+    Node names that had at least one C(Pulled image) log entry.
+  type: list
+  elements: str
+  returned: always
+"""
+
+import glob
+import os
+import re
+
+import yaml
+from ansible.module_utils.basic import AnsibleModule
+
+LOG_PATTERN = re.compile(
+    r'msg="Pulled image: (?P<actual_uri>[^@\s]+)@(?P<id>sha256:[a-f0-9]+)"'
+)
+SHA256_PATTERN = re.compile(r"sha256:[a-f0-9]+")
+
+
+def _node_from_path(path):
+    """Derive node name from the ``<node>.crio.log`` naming convention."""
+    basename = os.path.basename(path)
+    suffix_pos = basename.find(".crio.log")
+    if suffix_pos > 0:
+        return basename[:suffix_pos]
+    return os.path.splitext(basename)[0]
+
+
+def _domain_from_uri(uri):
+    """Return the registry host (+ optional port) from an image URI."""
+    return uri.split("/")[0].strip()
+
+
+def _apply_evidence(img, actual_uri, evidence_node, trusted_mirrors):
+    """Set common verification fields on an image row that has log evidence."""
+    actual_domain = _domain_from_uri(actual_uri)
+    img["node_verified_image_origin"] = (
+        "mirror" if actual_domain in trusted_mirrors else "source"
+    )
+    img["log_evidence_uri"] = actual_uri
+    img["log_evidence_node"] = evidence_node
+    return actual_domain
+
+
+def _collect_log_evidence(paths, module):
+    """Parse CRI-O logs into per-node and global evidence dicts.
+
+    Returns:
+        per_node: ``{node_name: {sha256_digest: pull_uri}}``
+        global_evidence: ``{sha256_digest: (pull_uri, node_name)}``
+            (last writer wins across nodes for the global dict)
+    """
+    per_node = {}
+    global_evidence = {}
+    for path in paths:
+        node = _node_from_path(path)
+        node_ev = per_node.setdefault(node, {})
+        try:
+            with open(path, "r") as f:
+                for line in f:
+                    match = LOG_PATTERN.search(line)
+                    if match:
+                        digest = match.group("id")
+                        uri = match.group("actual_uri")
+                        node_ev[digest] = uri
+                        global_evidence[digest] = (uri, node)
+        except IOError as exc:
+            module.fail_json(
+                msg="Cannot read CRI-O log file {0}: {1}".format(path, str(exc))
+            )
+    return per_node, global_evidence
+
+
+def run_module():
+    module_args = dict(
+        report_path=dict(type="str", required=True),
+        output_path=dict(type="str", required=True),
+        log_paths=dict(type="list", required=False, elements="str", default=[]),
+        log_dir=dict(type="str", required=False),
+        log_glob=dict(type="str", required=False, default="*.crio.log"),
+    )
+
+    module = AnsibleModule(argument_spec=module_args, supports_check_mode=True)
+
+    report_path = module.params["report_path"]
+    output_path = module.params["output_path"]
+    log_paths = module.params["log_paths"] or []
+    log_dir = module.params["log_dir"]
+    log_glob = module.params["log_glob"]
+
+    paths = list(log_paths)
+    if log_dir:
+        paths.extend(sorted(glob.glob(os.path.join(log_dir, log_glob))))
+
+    if not paths:
+        module.fail_json(
+            msg="No CRI-O log files: set log_paths and/or log_dir with matching files."
+        )
+
+    try:
+        with open(report_path, "r") as f:
+            data = yaml.safe_load(f)
+    except IOError as exc:
+        module.fail_json(
+            msg="Cannot read report {0}: {1}".format(report_path, str(exc))
+        )
+    except yaml.YAMLError as exc:
+        module.fail_json(msg="Invalid YAML in report: {0}".format(str(exc)))
+
+    if not isinstance(data, dict):
+        module.fail_json(msg="Report root must be a mapping (dict).")
+
+    trusted_mirrors = set()
+    summary_section = data.get("summary") or {}
+    for rule in summary_section.get("mirror_rules") or []:
+        if not isinstance(rule, dict):
+            continue
+        mirror_url = rule.get("mirror") or ""
+        if mirror_url:
+            domain = _domain_from_uri(mirror_url)
+            if domain:
+                trusted_mirrors.add(domain)
+
+    per_node_evidence, global_evidence = _collect_log_evidence(paths, module)
+
+    images_list = data.get("images") or []
+    entries_with_digest = 0
+    cross_node_entries = 0
+    for img in images_list:
+        if not isinstance(img, dict):
+            continue
+        image_id = img.get("image_id") or ""
+        sha_match = SHA256_PATTERN.search(image_id)
+        if not sha_match:
+            continue
+        entries_with_digest += 1
+        img_sha = sha_match.group(0)
+        img_node = img.get("node") or ""
+
+        node_local_hit = (
+            img_node
+            and img_node in per_node_evidence
+            and img_sha in per_node_evidence[img_node]
+        )
+
+        if node_local_hit:
+            actual_uri = per_node_evidence[img_node][img_sha]
+            actual_domain = _apply_evidence(img, actual_uri, img_node, trusted_mirrors)
+            expected_domain = img.get("expected_pull_location") or ""
+            img["verification_status"] = (
+                "MATCH" if actual_domain == expected_domain else "MISMATCH"
+            )
+        elif img_sha in global_evidence:
+            actual_uri, evidence_node = global_evidence[img_sha]
+            actual_domain = _apply_evidence(
+                img, actual_uri, evidence_node, trusted_mirrors
+            )
+            expected_domain = img.get("expected_pull_location") or ""
+            if img_node:
+                img["verification_status"] = "MISMATCH_CROSS_NODE"
+                cross_node_entries += 1
+            else:
+                img["verification_status"] = (
+                    "MATCH" if actual_domain == expected_domain else "MISMATCH"
+                )
+        else:
+            img["node_verified_image_origin"] = "cached/unknown"
+            img["log_evidence_uri"] = None
+            img["log_evidence_node"] = None
+            img["verification_status"] = "NOT_FOUND_IN_LOGS"
+
+    nodes_with_evidence = sorted(n for n, ev in per_node_evidence.items() if ev)
+    result = dict(
+        changed=False,
+        trusted_mirrors=sorted(trusted_mirrors),
+        log_files=len(paths),
+        entries_with_digest=entries_with_digest,
+        cross_node_entries=cross_node_entries,
+        nodes_with_evidence=nodes_with_evidence,
+    )
+
+    if module.check_mode:
+        result["changed"] = True
+        module.exit_json(**result)
+
+    try:
+        with open(output_path, "w") as f:
+            yaml.dump(data, f, default_flow_style=False, sort_keys=False)
+    except IOError as exc:
+        module.fail_json(
+            msg="Cannot write verified report {0}: {1}".format(output_path, str(exc))
+        )
+
+    result["changed"] = True
+    module.exit_json(**result)
+
+
+def main():
+    run_module()
+
+
+if __name__ == "__main__":
+    main()

roles/env_op_images/README.md

@@ -4,6 +4,9 @@ A role to gather the container images used in the openstack deployment with spec
 ## Parameters
 * `cifmw_env_op_images_dir`: (String) Directory where the operator_images.yaml will be stored. Defaults to `~/ci-framework-data/artifacts`
 * `cifmw_env_op_images_file`: (String) Name of the file storing the operator images and tags. Defaults to `operator_images.yaml`
+* `cifmw_env_op_images_pulled_report_file` / `cifmw_env_op_images_pulled_report_path`: Pulled-images policy report (ICSP/IDMS + pod image refs).
+* `cifmw_env_op_images_verified_report_file` / `cifmw_env_op_images_verified_report_path`: Output path for the CRI-O-enriched report. After the pulled report runs, fetches `oc adm node-logs NODE -u crio` per node, then writes this file with digest-level CRI-O fields (`node_verified_image_origin`, `log_evidence_uri`, `verification_status`).
+* `cifmw_env_op_images_crio_logs_dir`: Directory for per-node `*.crio.log` files used during verification.
 
 ## Examples
 ```YAML

roles/env_op_images/defaults/main.yml

@@ -23,6 +23,21 @@ cifmw_env_op_images_file: operator_images.yaml
 cifmw_env_op_images_dryrun: false
 
 cifmw_env_op_images_pulled_report_file: pulled_images_report.yaml
+cifmw_env_op_images_pulled_report_path: >-
+  {{
+    (cifmw_env_op_images_dir, 'artifacts', cifmw_env_op_images_pulled_report_file)
+    | path_join
+  }}
+
+cifmw_env_op_images_verified_report_file: pulled_images_report_verified.yaml
+cifmw_env_op_images_verified_report_path: >-
+  {{
+    (cifmw_env_op_images_dir, 'artifacts', cifmw_env_op_images_verified_report_file)
+    | path_join
+  }}
+cifmw_env_op_images_crio_logs_dir: >-
+  {{ (cifmw_env_op_images_dir, 'artifacts', 'crio_logs') | path_join }}
+
 cifmw_env_op_images_pulled_report_namespaces:
   - "{{ cifmw_openstack_namespace | default('openstack') }}"
   - "{{ operator_namespace | default('openstack-operators') }}"

roles/env_op_images/tasks/main.yml

@@ -158,3 +158,6 @@
 
 - name: Generate pulled images registry report
   ansible.builtin.include_tasks: pulled_images_report.yml
+
+- name: Verify pulled report against CRI-O node logs
+  ansible.builtin.include_tasks: verify_pulled_report_crio.yml

roles/env_op_images/tasks/pulled_images_report.yml

@@ -152,6 +152,7 @@
           {%     set _ = entries.append({
                    'namespace': pod.metadata.namespace,
                    'pod': pod.metadata.name,
+                   'node': pod.spec.get('nodeName', 'unknown'),
                    'container': cs.name,
                    'container_type': entry.type,
                    'operator': operator_name if operator_name else 'N/A',
@@ -192,11 +193,7 @@
           summary: "{{ _pulled_report_summary }}"
           images: "{{ _pulled_images_report }}"
       ansible.builtin.copy:
-        dest: >-
-          {{
-            (cifmw_env_op_images_dir, 'artifacts',
-             cifmw_env_op_images_pulled_report_file) | path_join
-          }}
+        dest: "{{ cifmw_env_op_images_pulled_report_path }}"
         content: "{{ _full_report | to_nice_yaml }}"
         mode: "0644"