Handle existing, locked machines at start up. Use thread-agnostic mac…

…hine deletion on reimage failures.

modules/machinery/az.py

@@ -69,6 +69,10 @@
 # This is hard cap of 4 given the maximum preemption chain length of 4
 MAX_CONCURRENT_VMSS_OPERATIONS = 4
 
+# These global lists will be used for maintaining lists of machines that failed during reimaging
+vms_absent_from_vmss = []
+vms_timed_out_being_reimaged = []
+
 # These global lists will be used for maintaining lists of ongoing operations on specific machines
 vms_currently_being_reimaged = []
 vms_currently_being_deleted = []
@@ -104,10 +108,6 @@ class Azure(Machinery):
     WINDOWS_PLATFORM = "windows"
     LINUX_PLATFORM = "linux"
 
-    # Statuses for machines
-    ABSENT = "absent_vm"
-    REIMAGE_FAILING = "failed_reimaging"
-
     def set_options(self, options: dict) -> None:
         """Set machine manager options.
         @param options: machine manager options dict.
@@ -118,11 +118,20 @@ def set_options(self, options: dict) -> None:
         if not isinstance(mmanager_opts["scale_sets"], list):
             mmanager_opts["scale_sets"] = str(mmanager_opts["scale_sets"]).strip().split(",")
 
+    def initialize(self):
+        """
+        Overloading abstracts.py:_initialize()
+        """
+        # Load.
+        self._initialize()
+
+        # Run initialization checks.
+        self._initialize_check()
+
     def _initialize(self):
         """
         Overloading abstracts.py:_initialize()
         Read configuration.
-        @param module_name: module name
         @raise CuckooDependencyError: if there is a problem with the dependencies call
         """
         mmanager_opts = self.options.get(self.module_name)
@@ -286,6 +295,7 @@ def _set_vmss_stage(self):
         self._process_pre_existing_vmsss()
         self._check_cpu_cores()
         self._get_or_upsert_vmsss(self.required_vmsss)
+        self._check_locked_machines()
         self._create_batch_threads()
 
     def _process_pre_existing_vmsss(self):
@@ -432,6 +442,17 @@ def _get_or_upsert_vmsss(self, vmsss_dict):
         for thr in vmss_reimage_threads + vmss_creation_threads:
             thr.join()
 
+    def _check_locked_machines(self):
+        """
+        In the case of CAPE unexpectedly restarting, release any locked machines.
+        They will have been reimaged and their tasks rescheduled before reaching this code.
+        """
+        running = self.running()
+        if len(running) > 0:
+            log.info("%d machines found locked on initialize, unlocking.", len(running))
+            for machine in running:
+                self.db.unlock_machine(machine)
+
     def _create_batch_threads(self):
         """
         Create batch reimage and delete threads.
@@ -491,9 +512,13 @@ def release(self, machine: Machine):
         @param label: machine label.
         """
         vmss_name = machine.label.split("_")[0]
-        if machine.status == Azure.ABSENT:
+        if machine.label in vms_absent_from_vmss:
             self.delete_machine(machine.label, delete_from_vmss=False)
-        elif machine.status == Azure.REIMAGE_FAILING or machine_pools[vmss_name]["is_scaling_down"]:
+            vms_absent_from_vmss.remove(machine.label)
+        elif machine.label in vms_timed_out_being_reimaged:
+            self.delete_machine(machine.label)
+            vms_timed_out_being_reimaged.remove(machine.label)
+        elif machine_pools[vmss_name]["is_scaling_down"]:
             self.delete_machine(machine.label)
         else:
             _ = super(Azure, self).release(machine)
@@ -1288,7 +1313,7 @@ def _thr_reimage_list_reader(self):
                                 vms_currently_being_deleted.append(f"{vmss_to_reimage}_{instance_id}")
                             with delete_lock:
                                 delete_vm_list.append({"vmss": vmss_to_reimage, "id": instance_id, "time_added": time.time()})
-                        self.set_status(f"{vmss_to_reimage}_{instance_id}", Azure.ABSENT)
+                        vms_absent_from_vmss.append(f"{vmss_to_reimage}_{instance_id}")
                         vms_currently_being_reimaged.remove(f"{vmss_to_reimage}_{instance_id}")
                         instance_ids.remove(instance_id)
 
@@ -1309,7 +1334,7 @@ def _thr_reimage_list_reader(self):
                         )
                         # That sucks, now we have mark each one for deletion
                         for instance_id in instance_ids:
-                            self.set_status(f"{vmss_to_reimage}_{instance_id}", Azure.REIMAGE_FAILING)
+                            vms_timed_out_being_reimaged.append(f"{vmss_to_reimage}_{instance_id}")
                         break
                     time.sleep(2)