Bug 2065 fixed
[osm/N2VC.git] / n2vc / k8s_helm_base_conn.py
index dd2116c..afb7bc2 100644 (file)
@@ -90,6 +90,9 @@ class K8sHelmBaseConnector(K8sConnector):
         if self._stable_repo_url == "None":
             self._stable_repo_url = None
 
+        # Lock to avoid concurrent execution of helm commands
+        self.cmd_lock = asyncio.Lock()
+
     def _get_namespace(self, cluster_uuid: str) -> str:
         """
         Obtains the namespace used by the cluster with the uuid passed by argument
@@ -401,9 +404,9 @@ class K8sHelmBaseConnector(K8sConnector):
         # version
         kdu_model, version = self._split_version(kdu_model)
 
-        repo = self._split_repo(kdu_model)
+        _, repo = self._split_repo(kdu_model)
         if repo:
-            self.repo_update(cluster_id, repo)
+            await self.repo_update(cluster_id, repo)
 
         command = self._get_install_command(
             kdu_model,
@@ -434,7 +437,6 @@ class K8sHelmBaseConnector(K8sConnector):
                     namespace=namespace,
                     db_dict=db_dict,
                     operation="install",
-                    run_once=False,
                 )
             )
 
@@ -463,8 +465,6 @@ class K8sHelmBaseConnector(K8sConnector):
             namespace=namespace,
             db_dict=db_dict,
             operation="install",
-            run_once=True,
-            check_every=0,
         )
 
         if rc != 0:
@@ -508,9 +508,9 @@ class K8sHelmBaseConnector(K8sConnector):
         # version
         kdu_model, version = self._split_version(kdu_model)
 
-        repo = self._split_repo(kdu_model)
+        _, repo = self._split_repo(kdu_model)
         if repo:
-            self.repo_update(cluster_uuid, repo)
+            await self.repo_update(cluster_uuid, repo)
 
         command = self._get_upgrade_command(
             kdu_model,
@@ -541,7 +541,6 @@ class K8sHelmBaseConnector(K8sConnector):
                     namespace=instance_info["namespace"],
                     db_dict=db_dict,
                     operation="upgrade",
-                    run_once=False,
                 )
             )
 
@@ -569,8 +568,6 @@ class K8sHelmBaseConnector(K8sConnector):
             namespace=instance_info["namespace"],
             db_dict=db_dict,
             operation="upgrade",
-            run_once=True,
-            check_every=0,
         )
 
         if rc != 0:
@@ -680,7 +677,6 @@ class K8sHelmBaseConnector(K8sConnector):
                     namespace=instance_info["namespace"],
                     db_dict=db_dict,
                     operation="scale",
-                    run_once=False,
                 )
             )
 
@@ -703,8 +699,6 @@ class K8sHelmBaseConnector(K8sConnector):
             namespace=instance_info["namespace"],
             db_dict=db_dict,
             operation="scale",
-            run_once=True,
-            check_every=0,
         )
 
         if rc != 0:
@@ -820,7 +814,6 @@ class K8sHelmBaseConnector(K8sConnector):
                 namespace=instance_info["namespace"],
                 db_dict=db_dict,
                 operation="rollback",
-                run_once=False,
             )
         )
 
@@ -839,8 +832,6 @@ class K8sHelmBaseConnector(K8sConnector):
             namespace=instance_info["namespace"],
             db_dict=db_dict,
             operation="rollback",
-            run_once=True,
-            check_every=0,
         )
 
         if rc != 0:
@@ -1518,17 +1509,18 @@ class K8sHelmBaseConnector(K8sConnector):
             environ.update(env)
 
         try:
-            process = await asyncio.create_subprocess_exec(
-                *command,
-                stdout=asyncio.subprocess.PIPE,
-                stderr=asyncio.subprocess.PIPE,
-                env=environ,
-            )
+            async with self.cmd_lock:
+                process = await asyncio.create_subprocess_exec(
+                    *command,
+                    stdout=asyncio.subprocess.PIPE,
+                    stderr=asyncio.subprocess.PIPE,
+                    env=environ,
+                )
 
-            # wait for command terminate
-            stdout, stderr = await process.communicate()
+                # wait for command terminate
+                stdout, stderr = await process.communicate()
 
-            return_code = process.returncode
+                return_code = process.returncode
 
             output = ""
             if stdout:
@@ -1555,6 +1547,9 @@ class K8sHelmBaseConnector(K8sConnector):
             return output, return_code
 
         except asyncio.CancelledError:
+            # first, kill the process if it is still running
+            if process.returncode is None:
+                process.kill()
             raise
         except K8sException:
             raise
@@ -1592,16 +1587,19 @@ class K8sHelmBaseConnector(K8sConnector):
             environ.update(env)
 
         try:
-            read, write = os.pipe()
-            await asyncio.create_subprocess_exec(*command1, stdout=write, env=environ)
-            os.close(write)
-            process_2 = await asyncio.create_subprocess_exec(
-                *command2, stdin=read, stdout=asyncio.subprocess.PIPE, env=environ
-            )
-            os.close(read)
-            stdout, stderr = await process_2.communicate()
+            async with self.cmd_lock:
+                read, write = os.pipe()
+                process_1 = await asyncio.create_subprocess_exec(
+                    *command1, stdout=write, env=environ
+                )
+                os.close(write)
+                process_2 = await asyncio.create_subprocess_exec(
+                    *command2, stdin=read, stdout=asyncio.subprocess.PIPE, env=environ
+                )
+                os.close(read)
+                stdout, stderr = await process_2.communicate()
 
-            return_code = process_2.returncode
+                return_code = process_2.returncode
 
             output = ""
             if stdout:
@@ -1627,6 +1625,10 @@ class K8sHelmBaseConnector(K8sConnector):
 
             return output, return_code
         except asyncio.CancelledError:
+            # first, kill the processes if they are still running
+            for process in (process_1, process_2):
+                if process.returncode is None:
+                    process.kill()
             raise
         except K8sException:
             raise
@@ -1713,10 +1715,8 @@ class K8sHelmBaseConnector(K8sConnector):
         if repo_url:
             repo_str = " --repo {}".format(repo_url)
 
-        idx = kdu_model.find("/")
-        if idx >= 0:
-            idx += 1
-            kdu_model = kdu_model[idx:]
+            # Obtain the Chart's name and store it in the var kdu_model
+            kdu_model, _ = self._split_repo(kdu_model=kdu_model)
 
         kdu_model, version = self._split_version(kdu_model)
         if version:
@@ -1725,10 +1725,13 @@ class K8sHelmBaseConnector(K8sConnector):
             version_str = ""
 
         full_command = self._get_inspect_command(
-            inspect_command, kdu_model, repo_str, version_str
+            show_command=inspect_command,
+            kdu_model=kdu_model,
+            repo_str=repo_str,
+            version=version_str,
         )
 
-        output, _rc = await self._local_async_exec(command=full_command)
+        output, _ = await self._local_async_exec(command=full_command)
 
         return output
 
@@ -1860,42 +1863,51 @@ class K8sHelmBaseConnector(K8sConnector):
         operation: str,
         kdu_instance: str,
         namespace: str = None,
-        check_every: float = 10,
         db_dict: dict = None,
-        run_once: bool = False,
-    ):
-        while True:
-            try:
-                await asyncio.sleep(check_every)
-                detailed_status = await self._status_kdu(
-                    cluster_id=cluster_id,
-                    kdu_instance=kdu_instance,
-                    yaml_format=False,
-                    namespace=namespace,
-                )
-                status = detailed_status.get("info").get("description")
-                self.log.debug("KDU {} STATUS: {}.".format(kdu_instance, status))
-                # write status to db
-                result = await self.write_app_status_to_db(
-                    db_dict=db_dict,
-                    status=str(status),
-                    detailed_status=str(detailed_status),
-                    operation=operation,
-                )
-                if not result:
-                    self.log.info("Error writing in database. Task exiting...")
-                    return
-            except asyncio.CancelledError:
-                self.log.debug("Task cancelled")
-                return
-            except Exception as e:
-                self.log.debug(
-                    "_store_status exception: {}".format(str(e)), exc_info=True
-                )
-                pass
-            finally:
-                if run_once:
-                    return
+    ) -> None:
+        """
+        Obtains the status of the KDU instance based on Helm Charts, and stores it in the database.
+
+        :param cluster_id (str): the cluster where the KDU instance is deployed
+        :param operation (str): The operation related to the status to be updated (for instance, "install" or "upgrade")
+        :param kdu_instance (str): The KDU instance in relation to which the status is obtained
+        :param namespace (str): The Kubernetes namespace where the KDU instance was deployed. Defaults to None
+        :param db_dict (dict): A dictionary with the database necessary information. It shall contain the
+        values for the keys:
+            - "collection": The Mongo DB collection to write to
+            - "filter": The query filter to use in the update process
+            - "path": The dot separated keys which targets the object to be updated
+        Defaults to None.
+        """
+
+        try:
+            detailed_status = await self._status_kdu(
+                cluster_id=cluster_id,
+                kdu_instance=kdu_instance,
+                yaml_format=False,
+                namespace=namespace,
+            )
+
+            status = detailed_status.get("info").get("description")
+            self.log.debug(f"Status for KDU {kdu_instance} obtained: {status}.")
+
+            # write status to db
+            result = await self.write_app_status_to_db(
+                db_dict=db_dict,
+                status=str(status),
+                detailed_status=str(detailed_status),
+                operation=operation,
+            )
+
+            if not result:
+                self.log.info("Error writing in database. Task exiting...")
+
+        except asyncio.CancelledError as e:
+            self.log.warning(
+                f"Exception in method {self._store_status.__name__} (task cancelled): {e}"
+            )
+        except Exception as e:
+            self.log.warning(f"Exception in method {self._store_status.__name__}: {e}")
 
     # params for use in -f file
     # returns values file option and filename (in order to delete it at the end)
@@ -1988,12 +2000,27 @@ class K8sHelmBaseConnector(K8sConnector):
                 kdu_model = parts[0]
         return kdu_model, version
 
-    async def _split_repo(self, kdu_model: str) -> str:
+    def _split_repo(self, kdu_model: str) -> (str, str):
+        """Obtain the Helm Chart's repository and Chart's names from the KDU model
+
+        Args:
+            kdu_model (str): Associated KDU model
+
+        Returns:
+            (str, str): Tuple with the Chart name in index 0, and the repo name
+                        in index 2; if there was a problem finding them, return None
+                        for both
+        """
+
+        chart_name = None
         repo_name = None
+
         idx = kdu_model.find("/")
         if idx >= 0:
+            chart_name = kdu_model[idx + 1 :]
             repo_name = kdu_model[:idx]
-        return repo_name
+
+        return chart_name, repo_name
 
     async def _find_repo(self, kdu_model: str, cluster_uuid: str) -> str:
         """Obtain the Helm repository for an Helm Chart
@@ -2006,12 +2033,15 @@ class K8sHelmBaseConnector(K8sConnector):
             str: the repository URL; if Helm Chart is a local one, the function returns None
         """
 
+        _, repo_name = self._split_repo(kdu_model=kdu_model)
+
         repo_url = None
-        idx = kdu_model.find("/")
-        if idx >= 0:
-            repo_name = kdu_model[:idx]
+        if repo_name:
             # Find repository link
             local_repo_list = await self.repo_list(cluster_uuid)
             for repo in local_repo_list:
-                repo_url = repo["url"] if repo["name"] == repo_name else None
+                if repo["name"] == repo_name:
+                    repo_url = repo["url"]
+                    break  # it is not necessary to continue the loop if the repo link was found...
+
         return repo_url