Merge pull request #2469 from Azure/release-2.6.0.0

Merge release 2.6.0.0 into master

azurelinuxagent/common/cgroupapi.py

@@ -28,6 +28,7 @@
 from azurelinuxagent.common.conf import get_agent_pid_file_path
 from azurelinuxagent.common.exception import CGroupsException, ExtensionErrorCodes, ExtensionError, ExtensionOperationError
 from azurelinuxagent.common.future import ustr
+from azurelinuxagent.common.osutil import systemd
 from azurelinuxagent.common.utils import fileutil, shellutil
 from azurelinuxagent.common.utils.extensionprocessutil import handle_process_completion, read_output, \
     TELEMETRY_MESSAGE_MAX_LEN
@@ -201,6 +202,24 @@ def get_process_cgroup_paths(self, process_id):
 
         return cpu_cgroup_path, memory_cgroup_path
 
+    def get_unit_cgroup_paths(self, unit_name):
+        """
+        Returns a tuple with the path of the cpu and memory cgroups for the given unit.
+        The values returned can be None if the controller is not mounted.
+        Ex: ControlGroup=/azure.slice/walinuxagent.service
+        controlgroup_path[1:] = azure.slice/walinuxagent.service
+        """
+        controlgroup_path = systemd.get_unit_property(unit_name, "ControlGroup")
+        cpu_mount_point, memory_mount_point = self.get_cgroup_mount_points()
+
+        cpu_cgroup_path = os.path.join(cpu_mount_point, controlgroup_path[1:]) \
+            if cpu_mount_point is not None else None
+
+        memory_cgroup_path = os.path.join(memory_mount_point, controlgroup_path[1:]) \
+            if memory_mount_point is not None else None
+
+        return cpu_cgroup_path, memory_cgroup_path
+
     @staticmethod
     def get_cgroup2_controllers():
         """
@@ -230,16 +249,16 @@ def _is_systemd_failure(scope_name, stderr):
         return unit_not_found in stderr or scope_name not in stderr
 
     @staticmethod
-    def get_extension_cgroup_name(extension_name):
+    def get_extension_slice_name(extension_name):
         # Since '-' is used as a separator in systemd unit names, we replace it with '_' to prevent side-effects.
-        return EXTENSION_SLICE_PREFIX + "-" + extension_name.replace('-', '_')
+        return EXTENSION_SLICE_PREFIX + "-" + extension_name.replace('-', '_') + ".slice"
 
     def start_extension_command(self, extension_name, command, cmd_name, timeout, shell, cwd, env, stdout, stderr, error_code=ExtensionErrorCodes.PluginUnknownFailure):
         scope = "{0}_{1}".format(cmd_name, uuid.uuid4())
-        extension_slice_name = self.get_extension_cgroup_name(extension_name)
+        extension_slice_name = self.get_extension_slice_name(extension_name)
         with self._systemd_run_commands_lock:
             process = subprocess.Popen(  # pylint: disable=W1509
-                "systemd-run --unit={0} --scope --slice={1}.slice {2}".format(scope, extension_slice_name, command),
+                "systemd-run --unit={0} --scope --slice={1} {2}".format(scope, extension_slice_name, command),
                 shell=shell,
                 cwd=cwd,
                 stdout=stdout,
@@ -255,7 +274,7 @@ def start_extension_command(self, extension_name, command, cmd_name, timeout, sh
         logger.info("Started extension in unit '{0}'", scope_name)
 
         try:
-            cgroup_relative_path = os.path.join('azure.slice/azure-vmextensions.slice', extension_slice_name + ".slice")
+            cgroup_relative_path = os.path.join('azure.slice/azure-vmextensions.slice', extension_slice_name)
 
             cpu_cgroup_mountpoint, _ = self.get_cgroup_mount_points()
 
@@ -264,6 +283,7 @@ def start_extension_command(self, extension_name, command, cmd_name, timeout, sh
             else:
                 cpu_cgroup_path = os.path.join(cpu_cgroup_mountpoint, cgroup_relative_path)
                 CGroupsTelemetry.track_cgroup(CpuCgroup(extension_name, cpu_cgroup_path))
+
         except IOError as e:
             if e.errno == 2:  # 'No such file or directory'
                 logger.info("The extension command already completed; will not track resource usage")

azurelinuxagent/common/cgroupstelemetry.py

@@ -24,7 +24,7 @@
 class CGroupsTelemetry(object):
     """
     """
-    _tracked = []
+    _tracked = {}
     _track_throttled_time = False
     _rlock = threading.RLock()
 
@@ -47,37 +47,37 @@ def track_cgroup(cgroup):
 
         with CGroupsTelemetry._rlock:
             if not CGroupsTelemetry.is_tracked(cgroup.path):
-                CGroupsTelemetry._tracked.append(cgroup)
+                CGroupsTelemetry._tracked[cgroup.path] = cgroup
                 logger.info("Started tracking cgroup {0}", cgroup)
 
     @staticmethod
     def is_tracked(path):
         """
         Returns true if the given item is in the list of tracked items
-        O(n) operation. But limited to few cgroup objects we have.
+        O(1) operation.
         """
         with CGroupsTelemetry._rlock:
-            for cgroup in CGroupsTelemetry._tracked:
-                if path == cgroup.path:
-                    return True
+            if path in CGroupsTelemetry._tracked:
+                return True
 
         return False
 
     @staticmethod
     def stop_tracking(cgroup):
         """
-        Stop tracking the cgroups for the given name
+        Stop tracking the cgroups for the given path
         """
         with CGroupsTelemetry._rlock:
-            CGroupsTelemetry._tracked.remove(cgroup)
-            logger.info("Stopped tracking cgroup {0}", cgroup)
+            if cgroup.path in CGroupsTelemetry._tracked:
+                CGroupsTelemetry._tracked.pop(cgroup.path)
+                logger.info("Stopped tracking cgroup {0}", cgroup)
 
     @staticmethod
     def poll_all_tracked():
         metrics = []
-
+        inactive_cgroups = []
         with CGroupsTelemetry._rlock:
-            for cgroup in CGroupsTelemetry._tracked[:]:
+            for cgroup in CGroupsTelemetry._tracked.values():
                 try:
                     metrics.extend(cgroup.get_tracked_metrics(track_throttled_time=CGroupsTelemetry._track_throttled_time))
                 except Exception as e:
@@ -89,12 +89,14 @@ def poll_all_tracked():
                         logger.periodic_warn(logger.EVERY_HOUR, '[PERIODIC] Could not collect metrics for cgroup '
                                                                 '{0}. Error : {1}'.format(cgroup.name, ustr(e)))
                 if not cgroup.is_active():
-                    CGroupsTelemetry.stop_tracking(cgroup)
+                    inactive_cgroups.append(cgroup)
+            for inactive_cgroup in inactive_cgroups:
+                CGroupsTelemetry.stop_tracking(inactive_cgroup)
 
         return metrics
 
     @staticmethod
     def reset():
         with CGroupsTelemetry._rlock:
-            CGroupsTelemetry._tracked *= 0  # emptying the list
+            CGroupsTelemetry._tracked.clear()  # emptying the dictionary
             CGroupsTelemetry._track_throttled_time = False

azurelinuxagent/common/conf.py

@@ -136,7 +136,7 @@ def load_conf_from_file(conf_file_path, conf=__conf__):
     "Debug.CgroupLogMetrics": False,
     "Debug.CgroupDisableOnProcessCheckFailure": True,
     "Debug.CgroupDisableOnQuotaCheckFailure": True,
-    "Debug.EnableFastTrack": False,
+    "Debug.EnableFastTrack": True,
 }
 
 
@@ -159,6 +159,8 @@ def load_conf_from_file(conf_file_path, conf=__conf__):
     "ResourceDisk.MountOptions": None,
     "ResourceDisk.Filesystem": "ext3",
     "AutoUpdate.GAFamily": "Prod",
+    "Debug.CgroupMonitorExpiryTime": "2022-01-31",
+    "Debug.CgroupMonitorExtensionName": "Microsoft.Azure.Monitor.AzureMonitorLinuxAgent",
 }
 
 
@@ -182,6 +184,8 @@ def load_conf_from_file(conf_file_path, conf=__conf__):
     # versions of the Agent.
     #
     "Debug.CgroupCheckPeriod": 300,
+    "Debug.AgentCpuQuota": 75,
+    "Debug.EtpCollectionPeriod": 300
 }
 
 
@@ -532,10 +536,44 @@ def get_cgroup_disable_on_quota_check_failure(conf=__conf__):
     """
     return conf.get_switch("Debug.CgroupDisableOnQuotaCheckFailure", True)
 
+
+def get_agent_cpu_quota(conf=__conf__):
+    """
+    CPU quota for the agent as a percentage of 1 CPU (100% == 1 CPU)
+
+    NOTE: This option is experimental and may be removed in later versions of the Agent.
+    """
+    return conf.get_int("Debug.AgentCpuQuota", 75)
+
+def get_cgroup_monitor_expiry_time (conf=__conf__):
+    """
+    cgroups monitoring disabled after expiry time
+
+    NOTE: This option is experimental and may be removed in later versions of the Agent.
+    """
+    return conf.get("Debug.CgroupMonitorExpiryTime", "2022-01-31")
+
+def get_cgroup_monitor_extension_name (conf=__conf__):
+    """
+    cgroups monitoring extension name
+
+    NOTE: This option is experimental and may be removed in later versions of the Agent.
+    """
+    return conf.get("Debug.CgroupMonitorExtensionName", "Microsoft.Azure.Monitor.AzureMonitorLinuxAgent")
+
+
 def get_enable_fast_track(conf=__conf__):
     """
     If True, the agent use FastTrack when retrieving goal states
 
     NOTE: This option is experimental and may be removed in later versions of the Agent.
     """
-    return conf.get_switch("Debug.EnableFastTrack", False)
+    return conf.get_switch("Debug.EnableFastTrack", True)
+
+
+def get_etp_collection_period(conf=__conf__):
+    """
+    Determines the frequency to perform ETP collection on extensions telemetry events.
+    NOTE: This option is experimental and may be removed in later versions of the Agent.
+    """
+    return conf.get_int("Debug.EtpCollectionPeriod", 300)

azurelinuxagent/common/event.py

@@ -124,6 +124,7 @@ class WALAEventOperation:
     Unknown = "Unknown"
     Upgrade = "Upgrade"
     Update = "Update"
+    VmSettings = "VmSettings"
 
 
 SHOULD_ENCODE_MESSAGE_LEN = 80

azurelinuxagent/common/exception.py

@@ -103,9 +103,21 @@ class ExtensionDownloadError(ExtensionError):
     """
 
 
-class ExtensionConfigError(ExtensionError):
+class ExtensionsGoalStateError(ExtensionError):
     """
-    Error raised when extension config file is malformed
+    Error raised when the ExtensionsGoalState is malformed
+    """
+
+
+class ExtensionsConfigError(ExtensionsGoalStateError):
+    """
+    Error raised when the ExtensionsConfig is malformed
+    """
+
+
+class VmSettingsError(ExtensionsGoalStateError):
+    """
+    Error raised when the VmSettings are malformed
     """
 
 

azurelinuxagent/common/logcollector_manifests.py

@@ -126,4 +126,4 @@
 
 echo,### Gathering Disk Info ###
 diskinfo,
-"""
+"""

azurelinuxagent/common/osutil/factory.py

@@ -81,7 +81,7 @@ def _get_osutil(distro_name, distro_code_name, distro_version, distro_full_name)
     if distro_name == "kali":
         return DebianOSBaseUtil()
 
-    if distro_name == "coreos" or distro_code_name == "coreos":
+    if distro_name in ("flatcar", "coreos") or distro_code_name in ("flatcar", "coreos"):
         return CoreOSUtil()
 
     if distro_name in ("suse", "sle_hpc", "sles", "opensuse"):
@@ -98,7 +98,7 @@ def _get_osutil(distro_name, distro_code_name, distro_version, distro_full_name)
 
         return DebianOSBaseUtil()
 
-    if distro_name in ("redhat", "rhel", "centos", "oracle", "almalinux"):
+    if distro_name in ("redhat", "rhel", "centos", "oracle", "almalinux", "cloudlinux"):
         if Version(distro_version) < Version("7"):
             return Redhat6xOSUtil()