[gNOI] Add support for gNOI Factory Reset

Author: VSuryaprasad-HCL

host_modules/gnoi_reset.py

@@ -0,0 +1,351 @@
+"""gNOI reset module which performs factory reset."""
+
+import json
+import logging
+import threading
+import time
+from host_modules import host_service
+from host_modules import infra_host
+
+MOD_NAME = "gnoi_reset"
+
+# We don't execute any boot install commands to the non-switch-linux switches
+# because they don't have boot count as the switch-linux switches do.
+EXECUTE_BOOT_INSTALL_COMMAND = ""
+GET_BOOT_INSTALL_VALUE_COMMAND = ""
+EXECUTE_CLEANUP_COMMAND = []
+
+# Timeout for SONiC Host Service to be killed during reboot. After executing the
+# reboot command, we will wait for 260 seconds for the reboot to complete, where
+# we expect that SONiC Host Service will be killed during this waiting period if
+# the reboot is successful. If this module is still alive after the waiting
+# period, we can conclude that the reboot has failed. Each container can take up
+# to 20 seconds to get killed. In total, there are 10 containers, and adding a
+# buffer of 1 minute brings up the delay value to be 260 seconds.
+REBOOT_TIMEOUT = 260
+
+EXECUTE_COLD_REBOOT_COMMAND = "sudo reboot"
+
+logger = logging.getLogger(__name__)
+
+
+class GnoiReset(host_service.HostModule):
+  """DBus endpoint that executes the factory reset and returns the reset
+
+  status and response.
+  """
+
+  def __init__(self, mod_name):
+    self.lock = threading.Lock()
+    self.is_reset_ongoing = False
+    self.reset_request = {}
+    self.reset_response = {}
+    super(GnoiReset, self).__init__(mod_name)
+
+  def populate_reset_response(
+      self,
+      reset_success=True,
+      factory_os_unsupported=False,
+      zero_fill_unsupported=False,
+      detail="",
+  ) -> tuple[int, str]:
+    """Populate the factory reset response.
+
+    Args:
+        reset_success: A boolean type variable to indicate whether the factory
+          reset succeeds or not.
+        factory_os_unsupported: A boolean type variable to indicate whether the
+          restoring to factory_os succeeds or not.
+        zero_fill_unsupported: A boolean type variable to indicate whether the
+          request to zero fill succeeds or not.
+        detail: A string indicates the detailed error message of the factory
+          reset if the error is not either factory_os_unsupported or
+          zero_fill_unsupported.
+
+    Returns:
+        A integer that indicates whether the factory reset succeeds or not,
+        and a json-style of StartResponse protobuf defined in reset.proto.
+        The integer value will be 0 if the factory reset succeeds, or 1 if
+        there is any failure happens.
+
+        Examples of the return value:
+            (0, dbus.String('{"reset_success": {}}'))
+            (1, dbus.String('{
+                    "reset_error": {
+                        "other": true,
+                        "detail": "Previous reset is ongoing."
+                    }
+                }')
+            )
+    """
+    self.lock.acquire()
+    self.reset_response = {}
+    if reset_success:
+      self.reset_response["reset_success"] = {}
+    else:
+      self.reset_response["reset_error"] = {}
+      if factory_os_unsupported:
+        self.reset_response["reset_error"]["factory_os_unsupported"] = True
+      elif zero_fill_unsupported:
+        self.reset_response["reset_error"]["zero_fill_unsupported"] = True
+      else:
+        self.reset_response["reset_error"]["other"] = True
+      self.reset_response["reset_error"]["detail"] = detail
+    response_data = json.dumps(self.reset_response)
+    self.lock.release()
+    return 0 if reset_success else 1, response_data
+
+  def execute_reboot(self) -> None:
+    """Execute cold reboot and log the error.
+
+    when the reboot fails.
+    """
+    rc, stdout, stderr = infra_host.InfraHost._run_command(
+        EXECUTE_COLD_REBOOT_COMMAND
+    )
+    if rc:
+      logger.error(
+          "%s: Cold reboot failed execution with stdout: %s, stderr: %s.",
+          MOD_NAME,
+          stdout,
+          stderr,
+      )
+      return
+
+    time.sleep(REBOOT_TIMEOUT)
+    return
+
+  def _check_reboot_in_progress(self) -> tuple[int, str]:
+    """Checks if reboot is already in progress.
+
+    Returns:
+        A integer that indicates whether the factory reset succeeds or not,
+        and a json-style of StartResponse protobuf defined in reset.proto.
+        The integer value will be 0 if the factory reset succeeds, or 1 if
+        there is any failure happens.
+
+        Examples of the return value:
+            (0, dbus.String('{"reset_success": {}}'))
+            (1, dbus.String('{
+                    "reset_error": {
+                        "other": true,
+                        "detail": "Previous reset is ongoing."
+                    }
+                }')
+            )
+    """
+    self.lock.acquire()
+    is_reset_ongoing = self.is_reset_ongoing
+    self.lock.release()
+
+    rc, stdout, stderr = infra_host.InfraHost._run_command(
+        GET_BOOT_INSTALL_VALUE_COMMAND
+    )
+    if rc or not stdout:
+      logger.error(
+          "%s: Failed to get boot install value with stdout: %s, stderr: %s",
+          MOD_NAME,
+          stdout,
+          stderr,
+      )
+      self.is_reset_ongoing = False
+      return self.populate_reset_response(
+          reset_success=False, detail="Failed to get the boot install value."
+      )
+
+    # Example of a valid google-specific platform stdout here is:
+    # ["regionselect=a", "bootcount=0", "bootinstall=0"].
+    boot_install = 0
+    try:
+      boot_install = int(stdout[2].split("=")[1])
+    except (ValueError, IndexError) as error:
+      return self.populate_reset_response(
+          reset_success=False,
+          detail="Failed to get the boot install value with error: %s."
+          % str(error),
+      )
+
+    # Return without issuing the reset if the previous reset is ongoing.
+    if is_reset_ongoing or boot_install != 0:
+      return self.populate_reset_response(
+        reset_success=False, detail="Previous reset is ongoing."
+      )
+
+    return 0, ""
+
+  def _parse_arguments(self, options) -> tuple[int, str]:
+    """Parses and validates the given arguments into a reset request.
+
+    Args:
+        options: A json-style string of StartRequest protobuf defined in
+          factory_reset/reset.proto.
+
+    Returns:
+        A integer that indicates whether the factory reset succeeds or not,
+        and a json-style of StartResponse protobuf defined in reset.proto.
+        The integer value will be 0 if the factory reset succeeds, or 1 if
+        there is any failure happens.
+
+        Examples of the return value:
+            (0, dbus.String('{"reset_success": {}}'))
+            (1, dbus.String('{
+                    "reset_error": {
+                        "other": true,
+                        "detail": "Previous reset is ongoing."
+                    }
+                }')
+            )
+    """
+    self.reset_request = {}
+    try:
+      self.reset_request = json.loads(options)
+    except ValueError:
+      return self.populate_reset_response(
+          reset_success=False,
+          detail=(
+              "Failed to parse json formatted factory reset request "
+              "into python dict."
+          ),
+      )
+
+    # Reject the request if zero_fill is set.
+    if "zeroFill" in self.reset_request and self.reset_request["zeroFill"]:
+      return self.populate_reset_response(
+          reset_success=False,
+          zero_fill_unsupported=True,
+          detail="zero_fill operation is currently unsupported.",
+      )
+
+    # Issue a warning if retain_certs is set.
+    if "retainCerts" in self.reset_request and self.reset_request["retainCerts"]:
+      logger.warning("%s: retain_certs is currently ignored.", MOD_NAME)
+
+    return 0, ""
+
+  def _cleanup_images(self) -> None:
+    """Cleans up the installed images, preparing for a factory reset."""
+    logger.info("Cleaning up install images.")
+    # Cleanup all install artifacts.
+    for command in EXECUTE_CLEANUP_COMMAND:
+      rc, stdout, stderr = infra_host.InfraHost._run_command(command)
+      if rc:
+        # Cleaning up artifacts is best effort, so continue on failure.
+        logger.warning(
+            "%s: Command %s execution failed with stdout: %s, stderr: %s.",
+            MOD_NAME,
+            command,
+            stdout,
+            stderr,
+        )
+
+  def _execute_reboot(self) -> tuple[int, str]:
+    """Performs a cold reboot, putting the switch into boot install mode.
+
+    Returns
+        A integer that indicates whether the factory reset succeeds or not,
+        and a json-style of StartResponse protobuf defined in reset.proto.
+        The integer value will be 0 if the factory reset succeeds, or 1 if
+        there is any failure happens.
+
+        Examples of the return value:
+            (0, dbus.String('{"reset_success": {}}'))
+            (1, dbus.String('{
+                    "reset_error": {
+                        "other": true,
+                        "detail": "Previous reset is ongoing."
+                    }
+                }')
+            )
+
+    Raises:
+        RuntimeError: An error occurred when starting a new thread.
+    """
+    # Issue the boot install command.
+    rc, stdout, stderr = infra_host.InfraHost._run_command(
+        EXECUTE_BOOT_INSTALL_COMMAND
+    )
+    if rc:
+      logger.error(
+          "%s: Boot count execution with stdout: %s, stderr: %s.",
+          MOD_NAME,
+          stdout,
+          stderr,
+      )
+      self.is_reset_ongoing = False
+      return self.populate_reset_response(
+          reset_success=False, detail="Boot count execution failed."
+      )
+
+    # Issue a cold reboot in a new thread and clear the reset response if
+    # the reboot succeeds.
+    try:
+      t = threading.Thread(target=self.execute_reboot)
+      t.start()
+    except RuntimeError as error:
+      self.is_reset_ongoing = False
+      return self.populate_reset_response(
+          reset_success=False,
+          detail="Failed to start thread to execute reboot.",
+      )
+
+    return 0, ""
+
+  @host_service.method(
+      host_service.bus_name(MOD_NAME), in_signature="as", out_signature="is"
+  )
+  def issue_reset(self, options) -> tuple[int, str]:
+    """Issues the factory reset by performing the following steps
+
+    sequentially:
+        1. Checks that there is no other reset requests ongoing.
+        2. Issues a bootcount command to the switch if it runs switch-linux.
+        3. Issues the cold reboot command to the switch.
+
+    Args:
+        options: A json-style string of StartRequest protobuf defined in
+          factory_reset/reset.proto.
+
+    Returns:
+        A integer that indicates whether the factory reset succeeds or not,
+        and a json-style of StartResponse protobuf defined in reset.proto.
+        The integer value will be 0 always regardless of success or failure
+        to ensure that the FE consumes the response correctly.
+
+        Examples of the return value:
+            (0, dbus.String('{"reset_success": {}}'))
+            (0, dbus.String('{
+                    "reset_error": {
+                        "other": true,
+                        "detail": "Previous reset is ongoing."
+                    }
+                }')
+            )
+
+    Raises:
+        RuntimeError: An error occurred when starting a new thread.
+    """
+    # Override the error code to always note success, so that the FE consumes
+    # the response correctly.
+    print("Issueing reset from Back end")
+    rc, resp = self._parse_arguments(options)
+    if rc:
+      return 0, resp
+
+    rc, resp = self._check_reboot_in_progress()
+    if rc:
+      return 0, resp
+
+    self.is_reset_ongoing = True
+    if "factoryOs" in self.reset_request and self.reset_request["factoryOs"]:
+      self._cleanup_images()
+
+    rc, resp = self._execute_reboot()
+    if rc:
+      return 0, resp
+
+    return 0, self.populate_reset_response()[1]
+
+
+def register():
+  """Return the class name"""
+  return GnoiReset, MOD_NAME