Clear idrac jobs earlier in enroll and optionally reset drac completely

Author: stevekeay

python/understack-workflows/tests/test_enroll_server.py

@@ -289,24 +289,24 @@ def test_enrol_happy_path_uses_virtual_media_inspect_and_flips_back(mocker):
             ),
             call(
                 created_node.uuid,
-                "inspect",  # OOB redfish inspect for bios_name / basic info
-                cleansteps=None,
+                "clean",
+                cleansteps=[{"interface": "management", "step": "clear_job_queue"}],
                 runbook=None,
-                disable_ramdisk=None,
+                disable_ramdisk=True,
             ),
             call(
                 created_node.uuid,
-                "inspect",  # agent inspect via virtual media
+                "inspect",  # OOB redfish inspect for bios_name / basic info
                 cleansteps=None,
                 runbook=None,
                 disable_ramdisk=None,
             ),
             call(
                 created_node.uuid,
-                "clean",
-                cleansteps=[{"interface": "management", "step": "clear_job_queue"}],
+                "inspect",  # agent inspect
+                cleansteps=None,
                 runbook=None,
-                disable_ramdisk=True,
+                disable_ramdisk=None,
             ),
             call(
                 created_node.uuid,
@@ -421,24 +421,24 @@ def test_enrol_existing_failed_node_recovers_and_updates(mocker):
             ),
             call(
                 existing_node.uuid,
-                "inspect",  # OOB inspect
-                cleansteps=None,
+                "clean",
+                cleansteps=[{"interface": "management", "step": "clear_job_queue"}],
                 runbook=None,
-                disable_ramdisk=None,
+                disable_ramdisk=True,
             ),
             call(
                 existing_node.uuid,
-                "inspect",  # Agent inspect via virtual media
+                "inspect",  # OOB inspect
                 cleansteps=None,
                 runbook=None,
                 disable_ramdisk=None,
             ),
             call(
                 existing_node.uuid,
-                "clean",
-                cleansteps=[{"interface": "management", "step": "clear_job_queue"}],
+                "inspect",  # Agent inspect
+                cleansteps=None,
                 runbook=None,
-                disable_ramdisk=True,
+                disable_ramdisk=None,
             ),
             call(
                 existing_node.uuid,

python/understack-workflows/understack_workflows/ironic_node.py

@@ -173,6 +173,17 @@ def clear_pending_idrac_jobs(node: Node):
     )
 
 
+def reset_idrac_to_known_good_state(node: Node):
+    logger.info("%s performing known_good_state clean step", node.uuid)
+    transition(
+        node,
+        target_state="clean",
+        expected_state="manageable",
+        clean_steps=[{"interface": "management", "step": "known_good_state"}],
+        disable_ramdisk=True,
+    )
+
+
 def _driver_for(manufacturer: str) -> tuple[str, str]:
     """Answer the (driver, inspect_interface) for this server."""
     if manufacturer.startswith("Dell"):

python/understack-workflows/understack_workflows/main/enroll_server.py

@@ -65,6 +65,7 @@ def main() -> None:
         firmware_update=args.firmware_update,
         raid_configure=args.raid_configure,
         external_cmdb_id=args.external_cmdb_id,
+        reset_idrac=args.reset_idrac,
     )
 
 
@@ -74,6 +75,7 @@ def enroll(
     raid_configure: bool,
     old_password: str | None,
     external_cmdb_id: str | None = None,
+    reset_idrac: bool = False,
 ) -> None:
     logger.info("Starting enroll workflow for bmc_ip_address=%s", ip_address)
 
@@ -90,6 +92,13 @@ def enroll(
         external_cmdb_id=external_cmdb_id,
     )
 
+    # Clear stale iDRAC jobs before virtual-media inspection, or optionally
+    # reset the controller to a broader known-good state.
+    if reset_idrac:
+        ironic_node.reset_idrac_to_known_good_state(node)
+    else:
+        ironic_node.clear_pending_idrac_jobs(node)
+
     # Out-of-band redfish inspection populates data including baremetal ports.
     #
     # Our hooks augment the ironic baremetal port with the BMC-reported
@@ -120,9 +129,6 @@ def enroll(
         )
     logger.info("[node:%s] Selected PXE interface %s", node.uuid, pxe_interface)
 
-    # Clear the job queue - stale jobs can conflict with the ones we create:
-    ironic_node.clear_pending_idrac_jobs(node)
-
     # This sets the boot device to use for all future HTTP boots:
     if update_dell_bios_settings(bmc, pxe_interface=pxe_interface):
         logger.info("%s performing second inspection write BIOS settings", node.uuid)
@@ -278,6 +284,12 @@ def argument_parser():
         default=True,
         help="Configure RAID before inspection",
     )
+    parser.add_argument(
+        "--reset-idrac",
+        type=parse_bool,
+        default=False,
+        help="Reset iDRAC to known_good_state instead of clear_job_queue",
+    )
     parser.add_argument(
         "--external-cmdb-id",
         required=False,