Merge branch 'develop' into fix-rsm-update

azurelinuxagent/common/osutil/redhat.py

@@ -117,11 +117,39 @@ def set_hostname(self, hostname):
             logger.warn("[{0}] failed, attempting fallback".format(' '.join(hostnamectl_cmd)))
             DefaultOSUtil.set_hostname(self, hostname)
 
+    def get_nm_controlled(self):
+        ifname = self.get_if_name()
+        filepath = "/etc/sysconfig/network-scripts/ifcfg-{0}".format(ifname)
+        nm_controlled_cmd = ['grep', 'NM_CONTROLLED=', filepath]
+        try:
+            result = shellutil.run_command(nm_controlled_cmd, log_error=False, encode_output=False).rstrip()
+
+            if result and len(result.split('=')) > 1:
+                # Remove trailing white space and ' or " characters
+                value = result.split('=')[1].replace("'", '').replace('"', '').rstrip()
+                if value == "n" or value == "no":
+                    return False
+        except shellutil.CommandError as e:
+            # Command might fail because NM_CONTROLLED value is not in interface config file (exit code 1).
+            # Log warning for any other exit code.
+            # NM_CONTROLLED=y by default if not specified.
+            if e.returncode != 1:
+                logger.warn("[{0}] failed: {1}.\nAgent will continue to publish hostname without NetworkManager restart".format(' '.join(nm_controlled_cmd), e))
+        except Exception as e:
+            logger.warn("Unexpected error while retrieving value of NM_CONTROLLED in {0}: {1}.\nAgent will continue to publish hostname without NetworkManager restart".format(filepath, e))
+
+        return True
+
     def publish_hostname(self, hostname):
         """
-        Restart NetworkManager first before publishing hostname
+        Restart NetworkManager first before publishing hostname, only if the network interface is not controlled by the
+        NetworkManager service (as determined by NM_CONTROLLED=n in the interface configuration). If the NetworkManager
+        service is restarted before the agent publishes the hostname, and NM_controlled=y, a race condition may happen
+        between the NetworkManager service and the Guest Agent making changes to the network interface configuration
+        simultaneously.
         """
-        shellutil.run("service NetworkManager restart")
+        if not self.get_nm_controlled():
+            shellutil.run("service NetworkManager restart")
         super(RedhatOSUtil, self).publish_hostname(hostname)
 
     def register_agent_service(self):

azurelinuxagent/common/utils/fileutil.py

@@ -99,13 +99,15 @@ def get_line_startingwith(prefix, filepath):
     return None
 
 
-def mkdir(dirpath, mode=None, owner=None):
+def mkdir(dirpath, mode=None, owner=None, reset_mode_and_owner=True):
     if not os.path.isdir(dirpath):
         os.makedirs(dirpath)
-    if mode is not None:
-        chmod(dirpath, mode)
-    if owner is not None:
-        chowner(dirpath, owner)
+        reset_mode_and_owner = True  # force setting the mode and owner
+    if reset_mode_and_owner:
+        if mode is not None:
+            chmod(dirpath, mode)
+        if owner is not None:
+            chowner(dirpath, owner)
 
 
 def chowner(path, owner):

azurelinuxagent/ga/exthandlers.py

@@ -1051,7 +1051,7 @@ def get_extension_full_name(self, extension=None):
 
     def __set_command_execution_log(self, extension, execution_log_max_size):
         try:
-            fileutil.mkdir(self.get_log_dir(), mode=0o755)
+            fileutil.mkdir(self.get_log_dir(), mode=0o755, reset_mode_and_owner=False)
         except IOError as e:
             self.logger.error(u"Failed to create extension log dir: {0}", e)
         else:

azurelinuxagent/pa/deprovision/default.py

@@ -158,7 +158,11 @@ def del_lib_dir_files(self, warnings, actions):  # pylint: disable=W0613
             'partition',
             'Protocol',
             'SharedConfig.xml',
-            'WireServerEndpoint'
+            'WireServerEndpoint',
+            'published_hostname',
+            'fast_track.json',
+            'initial_goal_state',
+            'rsm_update.json'
         ]
         known_files_glob = [
             'Extensions.*.xml',

azurelinuxagent/pa/provision/default.py

@@ -172,9 +172,11 @@ def check_provisioned_file(self):
         s = fileutil.read_file(ProvisionHandler.provisioned_file_path()).strip()
         if not self.osutil.is_current_instance_id(s):
             if len(s) > 0:
-                logger.warn("VM is provisioned, "
-                            "but the VM unique identifier has changed -- "
-                            "clearing cached state")
+                msg = "VM is provisioned, but the VM unique identifier has changed. This indicates the VM may be " \
+                      "created from an image that was not properly deprovisioned or generalized, which can result in " \
+                      "unexpected behavior from the guest agent -- clearing cached state"
+                logger.warn(msg)
+                self.report_event(msg)
                 from azurelinuxagent.pa.deprovision \
                     import get_deprovision_handler
                 deprovision_handler = get_deprovision_handler()

tests/ga/test_exthandlers.py

@@ -287,6 +287,28 @@ def test_command_extension_log_truncates_correctly(self, mock_log_dir):
         with open(log_file_path) as truncated_log_file:
             self.assertEqual(truncated_log_file.read(), "{second_line}\n".format(second_line=second_line))
 
+    def test_set_logger_should_not_reset_the_mode_of_the_log_directory(self):
+        ext_log_dir = os.path.join(self.tmp_dir, "log_directory")
+
+        with patch("azurelinuxagent.common.conf.get_ext_log_dir", return_value=ext_log_dir):
+            ext_handler = Extension(name='foo')
+            ext_handler.version = "1.2.3"
+            ext_handler_instance = ExtHandlerInstance(ext_handler=ext_handler, protocol=None)
+            ext_handler_log_dir = os.path.join(ext_log_dir, ext_handler.name)
+
+            # Double-check the initial mode
+            get_mode = lambda f: os.stat(f).st_mode & 0o777
+            mode = get_mode(ext_handler_log_dir)
+            if mode != 0o755:
+                raise Exception("The initial mode of the log directory should be 0o755, got 0{0:o}".format(mode))
+
+            new_mode = 0o700
+            os.chmod(ext_handler_log_dir, new_mode)
+            ext_handler_instance.set_logger()
+
+            mode = get_mode(ext_handler_log_dir)
+            self.assertEqual(new_mode, mode, "The mode of the log directory should not have changed")
+
     def test_it_should_report_the_message_in_the_hearbeat(self):
         def heartbeat_with_message():
             return {'code': 0, 'formattedMessage': {'lang': 'en-US', 'message': 'This is a heartbeat message'},

tests_e2e/orchestrator/lib/agent_junit.py

@@ -55,6 +55,9 @@ def _received_message(self, message: MessageBase) -> None:
             if "Unexpected error in AgentTestSuite" in message.message:
                 # Ignore these errors, they are already reported as AgentTestResultMessages
                 return
+            if "TestFailedException" in message.message:
+                # Ignore these errors, they are already reported as test failures
+                return
             # Change the suite name to "_Runbook_" for LISA messages in order to separate them
             # from actual test results.
             message.suite_full_name = "_Runbook_"

tests_e2e/pipeline/pipeline-cleanup.yml

@@ -13,7 +13,7 @@ parameters:
   - name: older_than
     displayName: Delete resources older than (use the syntax of the "date -d" command)
     type: string
-    default: 1 day ago
+    default: 12 hours ago
 
   - name: service_connections
     type: object
@@ -45,12 +45,12 @@ steps:
 
           rest_endpoint=$(az cloud show --query "endpoints.resourceManager" -o tsv)
           
+          pattern="${{ parameters.name_pattern }}"
+          
           az rest --method GET \
             --url "${rest_endpoint}/subscriptions/${subscription_id}/resourcegroups" \
             --url-parameters api-version=2021-04-01 \$expand=createdTime \
             --output json \
             --query value \
-          | jq --arg date "$date" '.[] | select (.createdTime < $date).name' \
-          | grep -i '${{ parameters.name_pattern }}' \
-          | xargs -l -t -r az group delete --no-wait -y -n \
-          || echo "No resource groups found to delete"
+          | jq --arg date "$date" '.[] | select (.createdTime < $date).name | match("'${pattern}'"; "g").string' \
+          | xargs -l -t -r az group delete --subscription "${subscription_id}" --no-wait -y -n

tests_e2e/pipeline/scripts/setup-agent.sh

@@ -18,7 +18,7 @@
 #
 
 #
-# Script to setup the agent VM for the Azure Pipelines agent pool; it simply installs the Azure CLI and the Docker Engine.
+# Script to setup the agent VM for the Azure Pipelines agent pool; it simply installs the Azure CLI, the Docker Engine and jq.
 #
 
 set -euox pipefail
@@ -48,3 +48,7 @@ sudo apt-get install -y docker-ce docker-ce-cli containerd.io docker-buildx-plug
 
 # Verify that Docker Engine is installed correctly by running the hello-world image.
 sudo docker run hello-world
+
+# Install jq; it is used by the cleanup pipeline to parse the JSON output of the Azure CLI
+sudo apt-get install -y jq
+

tests_e2e/test_suites/agent_update.yml

@@ -3,11 +3,11 @@
 # Self-update: If vm not enrolled into RSM, it will validate agent uses self-update to update to latest version published
 name: "AgentUpdate"
 tests:
-#  - "agent_update/rsm_update.py"  will enable this test once we have a new test version published
+#  - "agent_update/rsm_update.py"  TODO: will enable this test once we have a new test version published
    - "agent_update/self_update.py"
 images:
     - "random(endorsed, 10)"
-    - "random(endorsed-arm64, 2)"
+#    - "random(endorsed-arm64, 2)"  TODO: HGPA not deployed on some arm64 hosts(so agent stuck on Vmesttings calls as per contract) and will enable once HGPA deployed there
 locations: "AzureCloud:eastus2euap"
 owns_vm: true
 skip_on_clouds: