CRAYSAT-1817: Automate procedure of setting next boot device to disk

IM:CRAYSAT-1817
Reviewer: Ryan

CHANGELOG.md

@@ -50,6 +50,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
   in the `platform-services` stage.
 - Adding PG_NOT_DEEP_SCRUBBED in allowable checks excluded during ceph health check as it is
   ignorable. 
+- Automate the procedure of setting next boot device to disk before the management nodes are
+  powered off as part of the full-system shutdown.
 
 ### Fixed
 - Updated `sat bootsys` to increase the default management NCN shutdown timeout

sat/cli/bootsys/mgmt_power.py

@@ -31,6 +31,7 @@
 import socket
 import subprocess
 import sys
+import re
 
 import inflect
 from paramiko.ssh_exception import SSHException
@@ -260,6 +261,56 @@ def finish_shutdown(hosts, username, password, ncn_shutdown_timeout, ipmi_timeou
             sys.exit(0)
 
 
+def set_next_boot_device_to_disk(ssh_client, ncns):
+    """
+    Set the next boot device to a disk for a list of nodes via SSH.
+
+    This function connects to each node via SSH, retrieves the current boot order, identifies the first available disk,
+    and sets the next boot device to the specified disk.
+
+    Args:
+        ssh_client (paramiko.SSHClient): An active SSH client used to connect to the nodes.
+        ncns (list): A list of node names (NCNs) to configure.
+    """
+    command = 'efibootmgr'
+
+    for ncn in ncns:
+        try:
+            ssh_client.connect(ncn)
+        except (SSHException, socket.error) as err:
+            LOGGER.warning(f'Unable to connect to node {ncn}: {err}')
+            continue
+
+        try:
+            _, stdout, stderr = ssh_client.exec_command(command)
+            exit_code = stdout.channel.recv_exit_status()
+            if exit_code != 0:
+                LOGGER.warning(f'Unable to determine boot order of {ncn}, {command} exited with exit code {exit_code}')
+                continue
+
+            boot_order = stdout.read().decode()
+            match = re.search(r'^Boot([0-9A-Fa-f]{4})\*?\s.*UEFI OS', boot_order, re.MULTILINE)
+            if not match:
+                LOGGER.warning(f'No disk boot entries found for {ncn}')
+                continue
+
+            next_boot = match.group(1)
+
+            # Set the next boot device
+            next_boot_disk = f'efibootmgr -n {next_boot}'
+            _, stdout, stderr = ssh_client.exec_command(next_boot_disk)
+            exit_code = stdout.channel.recv_exit_status()
+            if exit_code != 0:
+                LOGGER.warning(f'Failed to set next boot device for {ncn}, {next_boot_disk} exited with exit code '
+                               f'{exit_code}')
+                continue
+
+            LOGGER.info(f'Successfully set next boot device to disk (Boot{next_boot}) for {ncn}')
+
+        finally:
+            ssh_client.close()
+
+
 def do_mgmt_shutdown_power(username, password, excluded_ncns, ncn_shutdown_timeout, ipmi_timeout):
     """Power off NCNs.
 
@@ -283,6 +334,7 @@ def do_mgmt_shutdown_power(username, password, excluded_ncns, ncn_shutdown_timeo
     # Shutdown workers
     worker_ncns = other_ncns_by_role.get('workers', [])
     if worker_ncns:
+        set_next_boot_device_to_disk(ssh_client, worker_ncns)
         try:
             with IPMIConsoleLogger(worker_ncns, username, password):
                 LOGGER.info(f'Shutting down worker NCNs: {", ".join(worker_ncns)}')
@@ -300,6 +352,8 @@ def do_mgmt_shutdown_power(username, password, excluded_ncns, ncn_shutdown_timeo
     # Shutdown managers (except ncn-m001)
     manager_ncns = other_ncns_by_role.get('managers', [])
     if manager_ncns:
+        set_next_boot_device_to_disk(ssh_client, ['ncn-m001'])
+        set_next_boot_device_to_disk(ssh_client, manager_ncns)
         try:
             with IPMIConsoleLogger(manager_ncns, username, password):
                 LOGGER.info(f'Shutting down manager NCNs: {", ".join(manager_ncns)}')
@@ -324,6 +378,8 @@ def do_mgmt_shutdown_power(username, password, excluded_ncns, ncn_shutdown_timeo
     # Freeze Ceph on storage nodes and then shutdown
     storage_ncns = other_ncns_by_role.get('storage', [])
     if storage_ncns:
+        set_next_boot_device_to_disk(ssh_client, storage_ncns)
+
         LOGGER.info(f'Freezing Ceph and shutting down storage NCNs: {", ".join(storage_ncns)}')
         try:
             do_ceph_freeze()

tests/cli/bootsys/test_mgmt_power.py

@@ -24,18 +24,24 @@
 """
 Tests for the sat.cli.bootsys.mgmt_power module.
 """
+import logging
 from argparse import Namespace
 import unittest
-from unittest.mock import MagicMock, patch
+from unittest.mock import MagicMock, patch, call
+from unittest import mock
 
 from paramiko.ssh_exception import SSHException, NoValidConnectionsError
 
 from sat.cli.bootsys.mgmt_power import (
     do_power_off_ncns,
     SSHAvailableWaiter,
     IPMIPowerStateWaiter,
+    do_mgmt_shutdown_power,
+    FatalBootsysError,
+    set_next_boot_device_to_disk
 )
 from sat.waiting import WaitingFailure
+from sat.cli.bootsys.platform import do_ceph_freeze, FatalPlatformError
 
 
 class TestSSHAvailableWaiter(unittest.TestCase):
@@ -177,3 +183,160 @@ def test_mgmt_ncns_skip_prompt_power_off(self):
         do_power_off_ncns(self.args)
         self.mock_prompt_continue.assert_not_called()
         self.mock_do_mgmt_shutdown_power.assert_called_once()
+
+
+class TestDoMgmtShutdownPower(unittest.TestCase):
+    """Tests for the do_mgmt_shutdown_power function."""
+
+    def setUp(self):
+        """Set up mocks."""
+        self.username = 'admin'
+        self.password = 'password'
+
+        self.mock_args = mock.Mock()
+        self.ncn_shutdown_timeout = 1200
+        self.ipmi_timeout = 60
+
+        self.mock_get_and_verify_ncn_groups = mock.patch(
+            'sat.cli.bootsys.mgmt_power.get_and_verify_ncn_groups').start()
+        self.mock_get_ssh_client = mock.patch(
+            'sat.cli.bootsys.mgmt_power.get_ssh_client').start()
+        self.mock_set_next_boot_device_to_disk = mock.patch(
+            'sat.cli.bootsys.mgmt_power.set_next_boot_device_to_disk').start()
+        self.mock_start_shutdown = mock.patch(
+            'sat.cli.bootsys.mgmt_power.start_shutdown').start()
+        self.mock_finish_shutdown = mock.patch(
+            'sat.cli.bootsys.mgmt_power.finish_shutdown').start()
+        self.mock_do_ceph_freeze = mock.patch(
+            'sat.cli.bootsys.mgmt_power.do_ceph_freeze').start()
+        self.mock_do_ceph_unmounts = mock.patch(
+            'sat.cli.bootsys.mgmt_power.do_ceph_unmounts').start()
+        self.mock_IPMIConsoleLogger = mock.patch(
+            'sat.cli.bootsys.mgmt_power.IPMIConsoleLogger').start()
+
+        self.mock_ssh_client = mock.Mock()
+        self.mock_get_ssh_client.return_value = self.mock_ssh_client
+
+        self.mock_other_ncns_by_role = {
+            'workers': ['ncn-w001', 'ncn-w002'],
+            'managers': ['ncn-m002', 'ncn-m003'],
+            'storage': ['ncn-s001', 'ncn-s002']
+        }
+        self.mock_get_and_verify_ncn_groups.return_value = self.mock_other_ncns_by_role
+
+    def tearDown(self):
+        mock.patch.stopall()
+
+    def test_failed_ncn_verification(self):
+        """Test do_mgmt_shutdown_power when NCN verification fails."""
+        bad_ncn_msg = 'Failed to identify members of the following NCN subrole(s): ...'
+        self.mock_get_and_verify_ncn_groups.side_effect = FatalBootsysError(bad_ncn_msg)
+        with self.assertRaises(SystemExit):
+            with self.assertLogs(level=logging.ERROR) as cm:
+                do_mgmt_shutdown_power(self.username, self.password, self.mock_args.excluded_ncns,
+                                       self.ncn_shutdown_timeout, self.ipmi_timeout)
+
+        expected_err = f'Not proceeding with NCN power off: {bad_ncn_msg}'
+        self.assertEqual(cm.records[0].message, expected_err)
+
+    def test_do_mgmt_shutdown_power_success(self):
+        """Test do_mgmt_shutdown_power when all steps are successful."""
+        with self.assertLogs(level=logging.INFO) as cm:
+            do_mgmt_shutdown_power(self.username, self.password, self.mock_args.excluded_ncns,
+                                   self.ncn_shutdown_timeout, self.ipmi_timeout)
+
+        # Assert calls for worker NCNs
+        self.mock_set_next_boot_device_to_disk.assert_any_call(self.mock_ssh_client, ['ncn-w001', 'ncn-w002'])
+        self.mock_start_shutdown.assert_any_call(['ncn-w001', 'ncn-w002'], self.mock_ssh_client)
+        self.mock_finish_shutdown.assert_any_call(['ncn-w001', 'ncn-w002'], self.username, self.password,
+                                                  self.ncn_shutdown_timeout, self.ipmi_timeout)
+
+        # Assert calls for manager NCNs
+        self.mock_set_next_boot_device_to_disk.assert_any_call(self.mock_ssh_client, ['ncn-m002', 'ncn-m003'])
+        self.mock_start_shutdown.assert_any_call(['ncn-m002', 'ncn-m003'], self.mock_ssh_client)
+        self.mock_finish_shutdown.assert_any_call(['ncn-m002', 'ncn-m003'], self.username, self.password,
+                                                  self.ncn_shutdown_timeout, self.ipmi_timeout)
+
+        # Assert calls for storage NCNs
+        self.mock_set_next_boot_device_to_disk.assert_any_call(self.mock_ssh_client, ['ncn-s001', 'ncn-s002'])
+        self.mock_do_ceph_freeze.assert_called_once()
+        self.mock_start_shutdown.assert_any_call(['ncn-s001', 'ncn-s002'], self.mock_ssh_client)
+        self.mock_finish_shutdown.assert_any_call(['ncn-s001', 'ncn-s002'], self.username, self.password,
+                                                  self.ncn_shutdown_timeout, self.ipmi_timeout)
+
+        # Assert call for Ceph unmount on ncn-m001
+        self.mock_do_ceph_unmounts.assert_called_once_with(self.mock_ssh_client, 'ncn-m001')
+
+        self.assertEqual(cm.records[0].message, 'Shutting down worker NCNs: ncn-w001, ncn-w002')
+        self.assertEqual(cm.records[1].message, 'Waiting up to 1200 seconds for worker NCNs to shut down...')
+        self.assertEqual(cm.records[2].message, 'Shutting down manager NCNs: ncn-m002, ncn-m003')
+        self.assertEqual(cm.records[3].message, 'Waiting up to 1200 seconds for manager NCNs to shut down...')
+        self.assertEqual(cm.records[4].message, 'Freezing Ceph and shutting down storage NCNs: ncn-s001, ncn-s002')
+        self.assertEqual(cm.records[5].message, 'Ceph freeze completed successfully on storage NCNs.')
+        self.assertEqual(cm.records[6].message, 'Waiting up to 1200 seconds for storage NCNs to shut down...')
+        self.assertEqual(cm.records[7].message, 'Shutdown and power off of storage NCNs: ncn-s001, ncn-s002')
+        self.assertEqual(cm.records[8].message, 'Shutdown and power off of all management NCNs complete.')
+
+    def test_do_mgmt_shutdown_power_with_fatal_error(self):
+        """Test do_mgmt_shutdown_power when a fatal error occurs."""
+        self.mock_do_ceph_freeze.side_effect = FatalPlatformError('Ceph freeze failed')
+        with self.assertLogs(level=logging.ERROR) as cm:
+            with self.assertRaises(SystemExit):
+                do_mgmt_shutdown_power(self.username, self.password, self.mock_args.excluded_ncns,
+                                       self.ncn_shutdown_timeout, self.ipmi_timeout)
+
+        self.mock_get_and_verify_ncn_groups.assert_called_once_with(self.mock_args.excluded_ncns.union({'ncn-m001'}))
+        self.assertEqual(cm.records[-1].message, 'Failed to freeze Ceph on storage NCNs: Ceph freeze failed')
+
+    def test_set_next_boot_device_to_disk_success(self):
+        """Test that the function sets the boot device to disk successfully"""
+        mock_ssh_client = MagicMock()
+        mock_stdout = MagicMock()
+        mock_stderr = MagicMock()
+        mock_stdout.channel.recv_exit_status.return_value = 0
+        mock_stdout.read.return_value = b"Boot0001* UEFI OS\nBoot0002* Other OS"
+
+        mock_ssh_client.exec_command.return_value = (None, mock_stdout, mock_stderr)
+
+        ncns = ['ncn-w001', 'ncn-s001']
+        set_next_boot_device_to_disk(mock_ssh_client, ncns)
+
+        expected_connect_calls = [call('ncn-w001'), call('ncn-s001')]
+        mock_ssh_client.connect.assert_has_calls(expected_connect_calls, any_order=True)
+
+        expected_exec_calls = [
+            call('efibootmgr'),
+            call('efibootmgr -n 0001'),
+            call('efibootmgr'),
+            call('efibootmgr -n 0001')
+        ]
+        mock_ssh_client.exec_command.assert_has_calls(expected_exec_calls, any_order=True)
+
+    def test_set_next_boot_device_to_disk_ssh_fail(self):
+        """Test that the function handles SSH connection failures"""
+        mock_ssh_client = MagicMock()
+        mock_ssh_client.connect.side_effect = SSHException('ssh failed')
+
+        with patch('sat.cli.bootsys.mgmt_power.LOGGER') as mock_logger:
+            ncns = ['ncn-w001']
+            set_next_boot_device_to_disk(mock_ssh_client, ncns)
+
+            mock_logger.warning.assert_called_with('Unable to connect to node ncn-w001: ssh failed')
+
+    def test_set_next_boot_device_to_disk_command_fail(self):
+        """Test that the function handles command execution failures"""
+        mock_ssh_client = MagicMock()
+        mock_stdout = MagicMock()
+        mock_stderr = MagicMock()
+        mock_stdout.channel.recv_exit_status.side_effect = [1, 0]
+        mock_stdout.read.return_value = b""
+
+        mock_ssh_client.exec_command.return_value = (None, mock_stdout, mock_stderr)
+
+        with patch('sat.cli.bootsys.mgmt_power.LOGGER') as mock_logger:
+            ncns = ['ncn-w001']
+            set_next_boot_device_to_disk(mock_ssh_client, ncns)
+
+            mock_logger.warning.assert_called_with(
+                'Unable to determine boot order of ncn-w001, efibootmgr exited with exit code 1'
+            )