Scope UNSTABLE end-of-run fail to testing stages only

Use a dedicated test_stage_unstable_flag written only by testing-stage
rescues (openstack_test, lb via openstack_test, egressip, cpms, csi,
conformance). The final ocp_testing play fails only when that flag
exists.

Soft verification checks and install workarounds keep using
stage_unstable_flag without triggering end-of-run failure, preserving
their previous job-color behavior.

Pair with ci-framework-jobs failure-message: UNSTABLE on monolithic
shiftstack jobs for yellow Zuul UI on test failures.

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

Author: tusharjadhav3302

collection/stages/roles/conformance_test/tasks/run_conformance_test.yml

@@ -104,7 +104,7 @@
     - name: Mark the openshift tests as UNSTABLE
       ansible.builtin.include_role:
         name: tools_stage_results
-        tasks_from: mark_stage_unstable.yml
+        tasks_from: mark_test_stage_unstable.yml
       vars:
         unstable_msg: >-
           The openshift test Suite failed.

collection/stages/roles/cpms_test/tasks/run_cpms_test.yml

@@ -28,7 +28,7 @@
     - name: Mark the CPMS {{ cpms_tests_type }} tests as UNSTABLE
       ansible.builtin.include_role:
         name: tools_stage_results
-        tasks_from: mark_stage_unstable.yml
+        tasks_from: mark_test_stage_unstable.yml
       vars:
         unstable_msg: >-
           The {{ cpms_test_name }} {{ cpms_tests_type }} test suite failed.

collection/stages/roles/csi_tests/tasks/run_csi_test.yml

@@ -28,7 +28,7 @@
     - name: Mark the openshift tests as UNSTABLE
       ansible.builtin.include_role:
         name: tools_stage_results
-        tasks_from: mark_stage_unstable.yml
+        tasks_from: mark_test_stage_unstable.yml
       vars:
         unstable_msg: >-
           The openshift test Suite failed.

collection/stages/roles/egressip_tests/tasks/run_egressip_tests.yml

@@ -93,7 +93,7 @@
     - name: Mark the egressip tests as UNSTABLE
       ansible.builtin.include_role:
         name: tools_stage_results
-        tasks_from: mark_stage_unstable.yml
+        tasks_from: mark_test_stage_unstable.yml
       vars:
         unstable_msg: >-
           The EgressIP test suite failed.

collection/stages/roles/openstack_test/tasks/run_openstack_test.yml

@@ -207,7 +207,7 @@
     - name: Mark the openshift tests as UNSTABLE
       ansible.builtin.include_role:
         name: tools_stage_results
-        tasks_from: mark_stage_unstable.yml
+        tasks_from: mark_test_stage_unstable.yml
       vars:
         unstable_msg: >-
           The openshift test Suite failed.

collection/tools/roles/tools_stage_results/tasks/fail_if_stage_unstable.yml

@@ -0,0 +1,28 @@
+---
+- name: Check whether any testing stage was marked UNSTABLE
+  ansible.builtin.stat:
+    path: "{{ test_stage_unstable_flag }}"
+  register: _test_stage_unstable_flag
+
+- name: Read accumulated UNSTABLE testing stage messages
+  ansible.builtin.set_fact:
+    _test_stage_unstable_messages: >-
+      {{ lookup('ansible.builtin.file', test_stage_unstable_flag)
+         | split('\n')
+         | map('trim')
+         | reject('equalto', '')
+         | list
+         | join('; ') }}
+  when: _test_stage_unstable_flag.stat.exists
+
+- name: Print accumulated UNSTABLE testing stage messages
+  ansible.builtin.debug:
+    msg: "{{ _test_stage_unstable_messages }}"
+  when: _test_stage_unstable_flag.stat.exists
+
+- name: Fail the playbook when any testing stage was marked UNSTABLE
+  ansible.builtin.fail:
+    msg: >-
+      One or more testing stages were marked UNSTABLE:
+      {{ _test_stage_unstable_messages }}
+  when: _test_stage_unstable_flag.stat.exists

collection/tools/roles/tools_stage_results/tasks/fail_if_test_stage_unstable.yml

@@ -0,0 +1,17 @@
+---
+- name: Set the test stage unstable var as true
+  ansible.builtin.set_fact:
+    test_stage_unstable: true
+    unstable_warning: "Warning! {{ unstable_msg }}"
+
+- name: Print the test stage UNSTABLE message
+  ansible.builtin.debug:
+    msg: "{{ unstable_warning }}"
+
+- name: Mark the test stage result as UNSTABLE
+  ansible.builtin.lineinfile:
+    path: "{{ test_stage_unstable_flag }}"
+    line: "{{ unstable_warning }}"
+    create: yes
+    mode: u=rw,g=rw,o=r
+  delegate_to: localhost

collection/tools/roles/tools_stage_results/tasks/mark_test_stage_unstable.yml

@@ -32,11 +32,13 @@ report_dir: "{{ artifacts_dir }}/test_results"
 
 # Sets the Ansible Controller Node machine's HOME environment variable
 controller_home_dir: "{{ lookup('ansible.builtin.env', 'HOME') }}"
-# The resources_file and stage_unstable_flag are on the Ansible control node's HOME.
-# stage_unstable_flag accumulates UNSTABLE stage messages during the run; ocp_testing.yaml
-# fails at the end if the file exists so Zuul/ci-framework see a non-zero exit.
+# The resources_file and stage_unstable flags are on the Ansible control node's HOME.
+# test_stage_unstable_flag is written only by testing-stage rescues; ocp_testing.yaml
+# fails at the end if it exists so Zuul can report UNSTABLE for test failures.
+# stage_unstable_flag is used by soft verification/install workarounds (no end-of-run fail).
 resources_file: "{{ controller_home_dir }}/artifacts/resources.yml"
 stage_unstable_flag: "{{ controller_home_dir }}/artifacts/stage_unstable_flag"
+test_stage_unstable_flag: "{{ controller_home_dir }}/artifacts/test_stage_unstable_flag"
 
 admin_cloud: "default"
 admin_env_file: "{{ osp_config_dir }}/{{ admin_cloud }}rc"

configs/global.yml

@@ -10,6 +10,11 @@
         path: "{{ stage_unstable_flag }}"
         state: absent
 
+    - name: Cleanup workspace flags - Remove the test_stage_unstable_flag if it exists
+      ansible.builtin.file:
+        path: "{{ test_stage_unstable_flag }}"
+        state: absent
+
     - name: Update ocp_deployment_topology with overrides if defined
       # Apply ocp_deployment_topology_override on top of the base topology
       # The base configuration comes from `configs/global.yml`
@@ -120,13 +125,13 @@
   ansible.builtin.import_playbook: plays/egressip_tests.yaml
   when: "'egressip_tests' in stages"
 
-- name: Fail if any stage was marked UNSTABLE
+- name: Fail if any testing stage was marked UNSTABLE
   hosts: localhost
   gather_facts: false
   vars_files:
     - "../configs/global.yml"
   tasks:
-    - name: Check accumulated UNSTABLE stages and fail the playbook
+    - name: Check accumulated UNSTABLE testing stages and fail the playbook
       ansible.builtin.include_role:
         name: tools_stage_results
-        tasks_from: fail_if_stage_unstable.yml
+        tasks_from: fail_if_test_stage_unstable.yml