Bug 1152, error if system cluster not properly initialized
[osm/LCM.git] / osm_lcm / vim_sdn.py
index ed25820..6a3deb5 100644 (file)
 # under the License.
 ##
 
-import asyncio
+import yaml
 import logging
 import logging.handlers
-import ROclient
-from lcm_utils import LcmException, LcmBase
+from osm_lcm import ROclient
+from osm_lcm.lcm_utils import LcmException, LcmBase, deep_get
+from n2vc.k8s_helm_conn import K8sHelmConnector
+from n2vc.k8s_juju_conn import K8sJujuConnector
+from n2vc.exceptions import K8sException, N2VCException
 from osm_common.dbbase import DbException
 from copy import deepcopy
 
@@ -29,9 +32,10 @@ __author__ = "Alfonso Tierno"
 
 class VimLcm(LcmBase):
     # values that are encrypted at vim config because they are passwords
-    vim_config_encrypted = ("admin_password", "nsx_password", "vcenter_password")
+    vim_config_encrypted = {"1.1": ("admin_password", "nsx_password", "vcenter_password"),
+                            "default": ("admin_password", "nsx_password", "vcenter_password", "vrops_password")}
 
-    def __init__(self, db, msg, fs, lcm_tasks, ro_config, loop):
+    def __init__(self, db, msg, fs, lcm_tasks, config, loop):
         """
         Init, Connect to database, filesystem storage, and messaging
         :param config: two level dictionary with configuration. Top level should contain 'database', 'storage',
@@ -41,25 +45,41 @@ class VimLcm(LcmBase):
         self.logger = logging.getLogger('lcm.vim')
         self.loop = loop
         self.lcm_tasks = lcm_tasks
-        self.ro_config = ro_config
+        self.ro_config = config["ro_config"]
 
         super().__init__(db, msg, fs, self.logger)
 
     async def create(self, vim_content, order_id):
+
+        # HA tasks and backward compatibility:
+        # If 'vim_content' does not include 'op_id', we a running a legacy NBI version.
+        # In such a case, HA is not supported by NBI, 'op_id' is None, and lock_HA() will do nothing.
+        # Register 'create' task here for related future HA operations
+        op_id = vim_content.pop('op_id', None)
+        if not self.lcm_tasks.lock_HA('vim', 'create', op_id):
+            return
+
         vim_id = vim_content["_id"]
-        vim_content.pop("op_id", None)
         logging_text = "Task vim_create={} ".format(vim_id)
         self.logger.debug(logging_text + "Enter")
+
         db_vim = None
         db_vim_update = {}
         exc = None
         RO_sdn_id = None
+        operationState_HA = ''
+        detailed_status_HA = ''
         try:
             step = "Getting vim-id='{}' from db".format(vim_id)
             db_vim = self.db.get_one("vim_accounts", {"_id": vim_id})
             if vim_content.get("config") and vim_content["config"].get("sdn-controller"):
                 step = "Getting sdn-controller-id='{}' from db".format(vim_content["config"]["sdn-controller"])
                 db_sdn = self.db.get_one("sdns", {"_id": vim_content["config"]["sdn-controller"]})
+
+                # If the VIM account has an associated SDN account, also
+                # wait for any previous tasks in process for the SDN
+                await self.lcm_tasks.waitfor_related_HA('sdn', 'ANY', db_sdn["_id"])
+
                 if db_sdn.get("_admin") and db_sdn["_admin"].get("deployed") and db_sdn["_admin"]["deployed"].get("RO"):
                     RO_sdn_id = db_sdn["_admin"]["deployed"]["RO"]
                 else:
@@ -105,7 +125,9 @@ class VimLcm(LcmBase):
                     del vim_account_RO["config"]["sdn-controller"]
                 if "sdn-port-mapping" in vim_account_RO["config"]:
                     del vim_account_RO["config"]["sdn-port-mapping"]
-                for p in self.vim_config_encrypted:
+                vim_config_encrypted_keys = self.vim_config_encrypted.get(schema_version) or \
+                    self.vim_config_encrypted.get("default")
+                for p in vim_config_encrypted_keys:
                     if vim_account_RO["config"].get(p):
                         vim_account_RO["config"][p] = self.db.decrypt(vim_account_RO["config"][p],
                                                                       schema_version=schema_version,
@@ -115,6 +137,9 @@ class VimLcm(LcmBase):
             db_vim_update["_admin.deployed.RO-account"] = desc["uuid"]
             db_vim_update["_admin.operationalState"] = "ENABLED"
             db_vim_update["_admin.detailed-status"] = "Done"
+            # Mark the VIM 'create' HA task as successful
+            operationState_HA = 'COMPLETED'
+            detailed_status_HA = 'Done'
 
             # await asyncio.sleep(15)   # TODO remove. This is for test
             self.logger.debug(logging_text + "Exit Ok VIM account created at RO_vim_account_id={}".format(desc["uuid"]))
@@ -130,52 +155,57 @@ class VimLcm(LcmBase):
             if exc and db_vim:
                 db_vim_update["_admin.operationalState"] = "ERROR"
                 db_vim_update["_admin.detailed-status"] = "ERROR {}: {}".format(step, exc)
+                # Mark the VIM 'create' HA task as erroneous
+                operationState_HA = 'FAILED'
+                detailed_status_HA = "ERROR {}: {}".format(step, exc)
             try:
                 if db_vim_update:
                     self.update_db_2("vim_accounts", vim_id, db_vim_update)
+                # Register the VIM 'create' HA task either
+                # succesful or erroneous, or do nothing (if legacy NBI)
+                self.lcm_tasks.register_HA('vim', 'create', op_id,
+                                           operationState=operationState_HA,
+                                           detailed_status=detailed_status_HA)
             except DbException as e:
                 self.logger.error(logging_text + "Cannot update database: {}".format(e))
 
             self.lcm_tasks.remove("vim_account", vim_id, order_id)
 
     async def edit(self, vim_content, order_id):
+
+        # HA tasks and backward compatibility:
+        # If 'vim_content' does not include 'op_id', we a running a legacy NBI version.
+        # In such a case, HA is not supported by NBI, and the HA check always returns True
+        op_id = vim_content.pop('op_id', None)
+        if not self.lcm_tasks.lock_HA('vim', 'edit', op_id):
+            return
+
         vim_id = vim_content["_id"]
-        vim_content.pop("op_id", None)
         logging_text = "Task vim_edit={} ".format(vim_id)
         self.logger.debug(logging_text + "Enter")
+
         db_vim = None
         exc = None
         RO_sdn_id = None
         RO_vim_id = None
         db_vim_update = {}
+        operationState_HA = ''
+        detailed_status_HA = ''
         step = "Getting vim-id='{}' from db".format(vim_id)
         try:
-            db_vim = self.db.get_one("vim_accounts", {"_id": vim_id})
+            # wait for any previous tasks in process
+            await self.lcm_tasks.waitfor_related_HA('vim', 'edit', op_id)
 
-            # look if previous tasks in process
-            task_name, task_dependency = self.lcm_tasks.lookfor_related("vim_account", vim_id, order_id)
-            if task_dependency:
-                step = "Waiting for related tasks to be completed: {}".format(task_name)
-                self.logger.debug(logging_text + step)
-                # TODO write this to database
-                _, pending = await asyncio.wait(task_dependency, timeout=3600)
-                if pending:
-                    raise LcmException("Timeout waiting related tasks to be completed")
+            db_vim = self.db.get_one("vim_accounts", {"_id": vim_id})
 
             if db_vim.get("_admin") and db_vim["_admin"].get("deployed") and db_vim["_admin"]["deployed"].get("RO"):
                 if vim_content.get("config") and vim_content["config"].get("sdn-controller"):
                     step = "Getting sdn-controller-id='{}' from db".format(vim_content["config"]["sdn-controller"])
                     db_sdn = self.db.get_one("sdns", {"_id": vim_content["config"]["sdn-controller"]})
 
-                    # look if previous tasks in process
-                    task_name, task_dependency = self.lcm_tasks.lookfor_related("sdn", db_sdn["_id"])
-                    if task_dependency:
-                        step = "Waiting for related tasks to be completed: {}".format(task_name)
-                        self.logger.debug(logging_text + step)
-                        # TODO write this to database
-                        _, pending = await asyncio.wait(task_dependency, timeout=3600)
-                        if pending:
-                            raise LcmException("Timeout waiting related tasks to be completed")
+                    # If the VIM account has an associated SDN account, also
+                    # wait for any previous tasks in process for the SDN
+                    await self.lcm_tasks.waitfor_related_HA('sdn', 'ANY', db_sdn["_id"])
 
                     if db_sdn.get("_admin") and db_sdn["_admin"].get("deployed") and db_sdn["_admin"]["deployed"].get(
                             "RO"):
@@ -223,7 +253,9 @@ class VimLcm(LcmBase):
                 if "config" in vim_content:
                     vim_account_RO["config"] = vim_content["config"]
                 if vim_content.get("config"):
-                    for p in self.vim_config_encrypted:
+                    vim_config_encrypted_keys = self.vim_config_encrypted.get(schema_version) or \
+                        self.vim_config_encrypted.get("default")
+                    for p in vim_config_encrypted_keys:
                         if vim_content["config"].get(p):
                             vim_account_RO["config"][p] = self.db.decrypt(vim_content["config"][p],
                                                                           schema_version=schema_version,
@@ -235,6 +267,9 @@ class VimLcm(LcmBase):
                 # vim_thread. RO will remove and relaunch a new thread for this vim_account
                 await RO.edit("vim_account", RO_vim_id, descriptor=vim_account_RO)
                 db_vim_update["_admin.operationalState"] = "ENABLED"
+                # Mark the VIM 'edit' HA task as successful
+                operationState_HA = 'COMPLETED'
+                detailed_status_HA = 'Done'
 
             self.logger.debug(logging_text + "Exit Ok RO_vim_id={}".format(RO_vim_id))
             return
@@ -249,22 +284,45 @@ class VimLcm(LcmBase):
             if exc and db_vim:
                 db_vim_update["_admin.operationalState"] = "ERROR"
                 db_vim_update["_admin.detailed-status"] = "ERROR {}: {}".format(step, exc)
+                # Mark the VIM 'edit' HA task as erroneous
+                operationState_HA = 'FAILED'
+                detailed_status_HA = "ERROR {}: {}".format(step, exc)
             try:
                 if db_vim_update:
                     self.update_db_2("vim_accounts", vim_id, db_vim_update)
+                # Register the VIM 'edit' HA task either
+                # succesful or erroneous, or do nothing (if legacy NBI)
+                self.lcm_tasks.register_HA('vim', 'edit', op_id,
+                                           operationState=operationState_HA,
+                                           detailed_status=detailed_status_HA)
             except DbException as e:
                 self.logger.error(logging_text + "Cannot update database: {}".format(e))
 
             self.lcm_tasks.remove("vim_account", vim_id, order_id)
 
-    async def delete(self, vim_id, order_id):
+    async def delete(self, vim_content, order_id):
+
+        # HA tasks and backward compatibility:
+        # If 'vim_content' does not include 'op_id', we a running a legacy NBI version.
+        # In such a case, HA is not supported by NBI, and the HA check always returns True
+        op_id = vim_content.pop('op_id', None)
+        if not self.lcm_tasks.lock_HA('vim', 'delete', op_id):
+            return
+
+        vim_id = vim_content["_id"]
         logging_text = "Task vim_delete={} ".format(vim_id)
         self.logger.debug(logging_text + "Enter")
+
         db_vim = None
         db_vim_update = {}
         exc = None
+        operationState_HA = ''
+        detailed_status_HA = ''
         step = "Getting vim from db"
         try:
+            # wait for any previous tasks in process
+            await self.lcm_tasks.waitfor_related_HA('vim', 'delete', op_id)
+
             db_vim = self.db.get_one("vim_accounts", {"_id": vim_id})
             if db_vim.get("_admin") and db_vim["_admin"].get("deployed") and db_vim["_admin"]["deployed"].get("RO"):
                 RO_vim_id = db_vim["_admin"]["deployed"]["RO"]
@@ -288,7 +346,7 @@ class VimLcm(LcmBase):
                         raise
             else:
                 # nothing to delete
-                self.logger.error(logging_text + "Nohing to remove at RO")
+                self.logger.error(logging_text + "Nothing to remove at RO")
             self.db.del_one("vim_accounts", {"_id": vim_id})
             db_vim = None
             self.logger.debug(logging_text + "Exit Ok")
@@ -305,9 +363,17 @@ class VimLcm(LcmBase):
             if exc and db_vim:
                 db_vim_update["_admin.operationalState"] = "ERROR"
                 db_vim_update["_admin.detailed-status"] = "ERROR {}: {}".format(step, exc)
+                # Mark the VIM 'delete' HA task as erroneous
+                operationState_HA = 'FAILED'
+                detailed_status_HA = "ERROR {}: {}".format(step, exc)
+                self.lcm_tasks.register_HA('vim', 'delete', op_id,
+                                           operationState=operationState_HA,
+                                           detailed_status=detailed_status_HA)
             try:
                 if db_vim and db_vim_update:
                     self.update_db_2("vim_accounts", vim_id, db_vim_update)
+                # If the VIM 'delete' HA task was succesful, the DB entry has been deleted,
+                # which means that there is nowhere to register this task, so do nothing here.
             except DbException as e:
                 self.logger.error(logging_text + "Cannot update database: {}".format(e))
             self.lcm_tasks.remove("vim_account", vim_id, order_id)
@@ -317,7 +383,7 @@ class WimLcm(LcmBase):
     # values that are encrypted at wim config because they are passwords
     wim_config_encrypted = ()
 
-    def __init__(self, db, msg, fs, lcm_tasks, ro_config, loop):
+    def __init__(self, db, msg, fs, lcm_tasks, config, loop):
         """
         Init, Connect to database, filesystem storage, and messaging
         :param config: two level dictionary with configuration. Top level should contain 'database', 'storage',
@@ -327,18 +393,28 @@ class WimLcm(LcmBase):
         self.logger = logging.getLogger('lcm.vim')
         self.loop = loop
         self.lcm_tasks = lcm_tasks
-        self.ro_config = ro_config
+        self.ro_config = config["ro_config"]
 
         super().__init__(db, msg, fs, self.logger)
 
     async def create(self, wim_content, order_id):
+
+        # HA tasks and backward compatibility:
+        # If 'wim_content' does not include 'op_id', we a running a legacy NBI version.
+        # In such a case, HA is not supported by NBI, 'op_id' is None, and lock_HA() will do nothing.
+        # Register 'create' task here for related future HA operations
+        op_id = wim_content.pop('op_id', None)
+        self.lcm_tasks.lock_HA('wim', 'create', op_id)
+
         wim_id = wim_content["_id"]
-        wim_content.pop("op_id", None)
         logging_text = "Task wim_create={} ".format(wim_id)
         self.logger.debug(logging_text + "Enter")
+
         db_wim = None
         db_wim_update = {}
         exc = None
+        operationState_HA = ''
+        detailed_status_HA = ''
         try:
             step = "Getting wim-id='{}' from db".format(wim_id)
             db_wim = self.db.get_one("wim_accounts", {"_id": wim_id})
@@ -388,6 +464,9 @@ class WimLcm(LcmBase):
             db_wim_update["_admin.deployed.RO-account"] = desc["uuid"]
             db_wim_update["_admin.operationalState"] = "ENABLED"
             db_wim_update["_admin.detailed-status"] = "Done"
+            # Mark the WIM 'create' HA task as successful
+            operationState_HA = 'COMPLETED'
+            detailed_status_HA = 'Done'
 
             self.logger.debug(logging_text + "Exit Ok WIM account created at RO_wim_account_id={}".format(desc["uuid"]))
             return
@@ -402,35 +481,46 @@ class WimLcm(LcmBase):
             if exc and db_wim:
                 db_wim_update["_admin.operationalState"] = "ERROR"
                 db_wim_update["_admin.detailed-status"] = "ERROR {}: {}".format(step, exc)
+                # Mark the WIM 'create' HA task as erroneous
+                operationState_HA = 'FAILED'
+                detailed_status_HA = "ERROR {}: {}".format(step, exc)
             try:
                 if db_wim_update:
                     self.update_db_2("wim_accounts", wim_id, db_wim_update)
+                # Register the WIM 'create' HA task either
+                # succesful or erroneous, or do nothing (if legacy NBI)
+                self.lcm_tasks.register_HA('wim', 'create', op_id,
+                                           operationState=operationState_HA,
+                                           detailed_status=detailed_status_HA)
             except DbException as e:
                 self.logger.error(logging_text + "Cannot update database: {}".format(e))
             self.lcm_tasks.remove("wim_account", wim_id, order_id)
 
     async def edit(self, wim_content, order_id):
+
+        # HA tasks and backward compatibility:
+        # If 'wim_content' does not include 'op_id', we a running a legacy NBI version.
+        # In such a case, HA is not supported by NBI, and the HA check always returns True
+        op_id = wim_content.pop('op_id', None)
+        if not self.lcm_tasks.lock_HA('wim', 'edit', op_id):
+            return
+
         wim_id = wim_content["_id"]
-        wim_content.pop("op_id", None)
         logging_text = "Task wim_edit={} ".format(wim_id)
         self.logger.debug(logging_text + "Enter")
+
         db_wim = None
         exc = None
         RO_wim_id = None
         db_wim_update = {}
         step = "Getting wim-id='{}' from db".format(wim_id)
+        operationState_HA = ''
+        detailed_status_HA = ''
         try:
-            db_wim = self.db.get_one("wim_accounts", {"_id": wim_id})
+            # wait for any previous tasks in process
+            await self.lcm_tasks.waitfor_related_HA('wim', 'edit', op_id)
 
-            # look if previous tasks in process
-            task_name, task_dependency = self.lcm_tasks.lookfor_related("wim_account", wim_id, order_id)
-            if task_dependency:
-                step = "Waiting for related tasks to be completed: {}".format(task_name)
-                self.logger.debug(logging_text + step)
-                # TODO write this to database
-                _, pending = await asyncio.wait(task_dependency, timeout=3600)
-                if pending:
-                    raise LcmException("Timeout waiting related tasks to be completed")
+            db_wim = self.db.get_one("wim_accounts", {"_id": wim_id})
 
             if db_wim.get("_admin") and db_wim["_admin"].get("deployed") and db_wim["_admin"]["deployed"].get("RO"):
 
@@ -481,6 +571,9 @@ class WimLcm(LcmBase):
                 # wim_thread. RO will remove and relaunch a new thread for this wim_account
                 await RO.edit("wim_account", RO_wim_id, descriptor=wim_account_RO)
                 db_wim_update["_admin.operationalState"] = "ENABLED"
+                # Mark the WIM 'edit' HA task as successful
+                operationState_HA = 'COMPLETED'
+                detailed_status_HA = 'Done'
 
             self.logger.debug(logging_text + "Exit Ok RO_wim_id={}".format(RO_wim_id))
             return
@@ -495,21 +588,44 @@ class WimLcm(LcmBase):
             if exc and db_wim:
                 db_wim_update["_admin.operationalState"] = "ERROR"
                 db_wim_update["_admin.detailed-status"] = "ERROR {}: {}".format(step, exc)
+                # Mark the WIM 'edit' HA task as erroneous
+                operationState_HA = 'FAILED'
+                detailed_status_HA = "ERROR {}: {}".format(step, exc)
             try:
                 if db_wim_update:
                     self.update_db_2("wim_accounts", wim_id, db_wim_update)
+                # Register the WIM 'edit' HA task either
+                # succesful or erroneous, or do nothing (if legacy NBI)
+                self.lcm_tasks.register_HA('wim', 'edit', op_id,
+                                           operationState=operationState_HA,
+                                           detailed_status=detailed_status_HA)
             except DbException as e:
                 self.logger.error(logging_text + "Cannot update database: {}".format(e))
             self.lcm_tasks.remove("wim_account", wim_id, order_id)
 
-    async def delete(self, wim_id, order_id):
+    async def delete(self, wim_content, order_id):
+
+        # HA tasks and backward compatibility:
+        # If 'vim_content' does not include 'op_id', we a running a legacy NBI version.
+        # In such a case, HA is not supported by NBI, and the HA check always returns True
+        op_id = wim_content.pop('op_id', None)
+        if not self.lcm_tasks.lock_HA('wim', 'delete', op_id):
+            return
+
+        wim_id = wim_content["_id"]
         logging_text = "Task wim_delete={} ".format(wim_id)
         self.logger.debug(logging_text + "Enter")
+
         db_wim = None
         db_wim_update = {}
         exc = None
         step = "Getting wim from db"
+        operationState_HA = ''
+        detailed_status_HA = ''
         try:
+            # wait for any previous tasks in process
+            await self.lcm_tasks.waitfor_related_HA('wim', 'delete', op_id)
+
             db_wim = self.db.get_one("wim_accounts", {"_id": wim_id})
             if db_wim.get("_admin") and db_wim["_admin"].get("deployed") and db_wim["_admin"]["deployed"].get("RO"):
                 RO_wim_id = db_wim["_admin"]["deployed"]["RO"]
@@ -550,9 +666,17 @@ class WimLcm(LcmBase):
             if exc and db_wim:
                 db_wim_update["_admin.operationalState"] = "ERROR"
                 db_wim_update["_admin.detailed-status"] = "ERROR {}: {}".format(step, exc)
+                # Mark the WIM 'delete' HA task as erroneous
+                operationState_HA = 'FAILED'
+                detailed_status_HA = "ERROR {}: {}".format(step, exc)
+                self.lcm_tasks.register_HA('wim', 'delete', op_id,
+                                           operationState=operationState_HA,
+                                           detailed_status=detailed_status_HA)
             try:
                 if db_wim and db_wim_update:
                     self.update_db_2("wim_accounts", wim_id, db_wim_update)
+                # If the WIM 'delete' HA task was succesful, the DB entry has been deleted,
+                # which means that there is nowhere to register this task, so do nothing here.
             except DbException as e:
                 self.logger.error(logging_text + "Cannot update database: {}".format(e))
             self.lcm_tasks.remove("wim_account", wim_id, order_id)
@@ -560,7 +684,7 @@ class WimLcm(LcmBase):
 
 class SdnLcm(LcmBase):
 
-    def __init__(self, db, msg, fs, lcm_tasks, ro_config, loop):
+    def __init__(self, db, msg, fs, lcm_tasks, config, loop):
         """
         Init, Connect to database, filesystem storage, and messaging
         :param config: two level dictionary with configuration. Top level should contain 'database', 'storage',
@@ -570,19 +694,29 @@ class SdnLcm(LcmBase):
         self.logger = logging.getLogger('lcm.sdn')
         self.loop = loop
         self.lcm_tasks = lcm_tasks
-        self.ro_config = ro_config
+        self.ro_config = config["ro_config"]
 
         super().__init__(db, msg, fs, self.logger)
 
     async def create(self, sdn_content, order_id):
+
+        # HA tasks and backward compatibility:
+        # If 'sdn_content' does not include 'op_id', we a running a legacy NBI version.
+        # In such a case, HA is not supported by NBI, 'op_id' is None, and lock_HA() will do nothing.
+        # Register 'create' task here for related future HA operations
+        op_id = sdn_content.pop('op_id', None)
+        self.lcm_tasks.lock_HA('sdn', 'create', op_id)
+
         sdn_id = sdn_content["_id"]
-        sdn_content.pop("op_id", None)
         logging_text = "Task sdn_create={} ".format(sdn_id)
         self.logger.debug(logging_text + "Enter")
+
         db_sdn = None
         db_sdn_update = {}
         RO_sdn_id = None
         exc = None
+        operationState_HA = ''
+        detailed_status_HA = ''
         try:
             step = "Getting sdn from db"
             db_sdn = self.db.get_one("sdns", {"_id": sdn_id})
@@ -607,6 +741,9 @@ class SdnLcm(LcmBase):
             db_sdn_update["_admin.deployed.RO"] = RO_sdn_id
             db_sdn_update["_admin.operationalState"] = "ENABLED"
             self.logger.debug(logging_text + "Exit Ok RO_sdn_id={}".format(RO_sdn_id))
+            # Mark the SDN 'create' HA task as successful
+            operationState_HA = 'COMPLETED'
+            detailed_status_HA = 'Done'
             return
 
         except (ROclient.ROClientException, DbException) as e:
@@ -619,23 +756,44 @@ class SdnLcm(LcmBase):
             if exc and db_sdn:
                 db_sdn_update["_admin.operationalState"] = "ERROR"
                 db_sdn_update["_admin.detailed-status"] = "ERROR {}: {}".format(step, exc)
+                # Mark the SDN 'create' HA task as erroneous
+                operationState_HA = 'FAILED'
+                detailed_status_HA = "ERROR {}: {}".format(step, exc)
             try:
                 if db_sdn and db_sdn_update:
                     self.update_db_2("sdns", sdn_id, db_sdn_update)
+                # Register the SDN 'create' HA task either
+                # succesful or erroneous, or do nothing (if legacy NBI)
+                self.lcm_tasks.register_HA('sdn', 'create', op_id,
+                                           operationState=operationState_HA,
+                                           detailed_status=detailed_status_HA)
             except DbException as e:
                 self.logger.error(logging_text + "Cannot update database: {}".format(e))
             self.lcm_tasks.remove("sdn", sdn_id, order_id)
 
     async def edit(self, sdn_content, order_id):
+
+        # HA tasks and backward compatibility:
+        # If 'sdn_content' does not include 'op_id', we a running a legacy NBI version.
+        # In such a case, HA is not supported by NBI, and the HA check always returns True
+        op_id = sdn_content.pop('op_id', None)
+        if not self.lcm_tasks.lock_HA('sdn', 'edit', op_id):
+            return
+
         sdn_id = sdn_content["_id"]
-        sdn_content.pop("op_id", None)
         logging_text = "Task sdn_edit={} ".format(sdn_id)
         self.logger.debug(logging_text + "Enter")
+
         db_sdn = None
         db_sdn_update = {}
         exc = None
+        operationState_HA = ''
+        detailed_status_HA = ''
         step = "Getting sdn from db"
         try:
+            # wait for any previous tasks in process
+            await self.lcm_tasks.waitfor_related_HA('sdn', 'edit', op_id)
+
             db_sdn = self.db.get_one("sdns", {"_id": sdn_id})
             RO_sdn_id = None
             if db_sdn.get("_admin") and db_sdn["_admin"].get("deployed") and db_sdn["_admin"]["deployed"].get("RO"):
@@ -653,6 +811,9 @@ class SdnLcm(LcmBase):
                 if sdn_RO:
                     await RO.edit("sdn", RO_sdn_id, descriptor=sdn_RO)
                 db_sdn_update["_admin.operationalState"] = "ENABLED"
+                # Mark the SDN 'edit' HA task as successful
+                operationState_HA = 'COMPLETED'
+                detailed_status_HA = 'Done'
 
             self.logger.debug(logging_text + "Exit Ok RO_sdn_id={}".format(RO_sdn_id))
             return
@@ -667,21 +828,44 @@ class SdnLcm(LcmBase):
             if exc and db_sdn:
                 db_sdn["_admin.operationalState"] = "ERROR"
                 db_sdn["_admin.detailed-status"] = "ERROR {}: {}".format(step, exc)
+                # Mark the SDN 'edit' HA task as erroneous
+                operationState_HA = 'FAILED'
+                detailed_status_HA = "ERROR {}: {}".format(step, exc)
             try:
                 if db_sdn_update:
                     self.update_db_2("sdns", sdn_id, db_sdn_update)
+                # Register the SDN 'edit' HA task either
+                # succesful or erroneous, or do nothing (if legacy NBI)
+                self.lcm_tasks.register_HA('sdn', 'edit', op_id,
+                                           operationState=operationState_HA,
+                                           detailed_status=detailed_status_HA)
             except DbException as e:
                 self.logger.error(logging_text + "Cannot update database: {}".format(e))
             self.lcm_tasks.remove("sdn", sdn_id, order_id)
 
-    async def delete(self, sdn_id, order_id):
+    async def delete(self, sdn_content, order_id):
+
+        # HA tasks and backward compatibility:
+        # If 'vim_content' does not include 'op_id', we a running a legacy NBI version.
+        # In such a case, HA is not supported by NBI, and the HA check always returns True
+        op_id = sdn_content.pop('op_id', None)
+        if not self.lcm_tasks.lock_HA('sdn', 'delete', op_id):
+            return
+
+        sdn_id = sdn_content["_id"]
         logging_text = "Task sdn_delete={} ".format(sdn_id)
         self.logger.debug(logging_text + "Enter")
+
         db_sdn = None
         db_sdn_update = {}
         exc = None
+        operationState_HA = ''
+        detailed_status_HA = ''
         step = "Getting sdn from db"
         try:
+            # wait for any previous tasks in process
+            await self.lcm_tasks.waitfor_related_HA('sdn', 'delete', op_id)
+
             db_sdn = self.db.get_one("sdns", {"_id": sdn_id})
             if db_sdn.get("_admin") and db_sdn["_admin"].get("deployed") and db_sdn["_admin"]["deployed"].get("RO"):
                 RO_sdn_id = db_sdn["_admin"]["deployed"]["RO"]
@@ -712,9 +896,351 @@ class SdnLcm(LcmBase):
             if exc and db_sdn:
                 db_sdn["_admin.operationalState"] = "ERROR"
                 db_sdn["_admin.detailed-status"] = "ERROR {}: {}".format(step, exc)
+                # Mark the SDN 'delete' HA task as erroneous
+                operationState_HA = 'FAILED'
+                detailed_status_HA = "ERROR {}: {}".format(step, exc)
+                self.lcm_tasks.register_HA('sdn', 'delete', op_id,
+                                           operationState=operationState_HA,
+                                           detailed_status=detailed_status_HA)
             try:
                 if db_sdn and db_sdn_update:
                     self.update_db_2("sdns", sdn_id, db_sdn_update)
+                # If the SDN 'delete' HA task was succesful, the DB entry has been deleted,
+                # which means that there is nowhere to register this task, so do nothing here.
             except DbException as e:
                 self.logger.error(logging_text + "Cannot update database: {}".format(e))
             self.lcm_tasks.remove("sdn", sdn_id, order_id)
+
+
+class K8sClusterLcm(LcmBase):
+
+    def __init__(self, db, msg, fs, lcm_tasks, config, loop):
+        """
+        Init, Connect to database, filesystem storage, and messaging
+        :param config: two level dictionary with configuration. Top level should contain 'database', 'storage',
+        :return: None
+        """
+
+        self.logger = logging.getLogger('lcm.k8scluster')
+        self.loop = loop
+        self.lcm_tasks = lcm_tasks
+        self.vca_config = config["VCA"]
+        self.fs = fs
+        self.db = db
+
+        self.helm_k8scluster = K8sHelmConnector(
+            kubectl_command=self.vca_config.get("kubectlpath"),
+            helm_command=self.vca_config.get("helmpath"),
+            fs=self.fs,
+            log=self.logger,
+            db=self.db,
+            on_update_db=None
+        )
+
+        self.juju_k8scluster = K8sJujuConnector(
+            kubectl_command=self.vca_config.get("kubectlpath"),
+            juju_command=self.vca_config.get("jujupath"),
+            fs=self.fs,
+            log=self.logger,
+            db=self.db,
+            on_update_db=None
+        )
+
+        super().__init__(db, msg, fs, self.logger)
+
+    async def create(self, k8scluster_content, order_id):
+
+        # HA tasks and backward compatibility:
+        # If 'vim_content' does not include 'op_id', we a running a legacy NBI version.
+        # In such a case, HA is not supported by NBI, 'op_id' is None, and lock_HA() will do nothing.
+        # Register 'create' task here for related future HA operations
+        op_id = k8scluster_content.pop('op_id', None)
+        if not self.lcm_tasks.lock_HA('k8scluster', 'create', op_id):
+            return
+
+        k8scluster_id = k8scluster_content["_id"]
+        logging_text = "Task k8scluster_create={} ".format(k8scluster_id)
+        self.logger.debug(logging_text + "Enter")
+
+        db_k8scluster = None
+        db_k8scluster_update = {}
+
+        exc = None
+        operationState_HA = ''
+        detailed_status_HA = ''
+        try:
+            step = "Getting k8scluster-id='{}' from db".format(k8scluster_id)
+            self.logger.debug(logging_text + step)
+            db_k8scluster = self.db.get_one("k8sclusters", {"_id": k8scluster_id})
+            self.db.encrypt_decrypt_fields(db_k8scluster.get("credentials"), 'decrypt', ['password', 'secret'],
+                                           schema_version=db_k8scluster["schema_version"], salt=db_k8scluster["_id"])
+            k8s_credentials = yaml.safe_dump(db_k8scluster.get("credentials"))
+            error_text_list = []
+            init_target = deep_get(db_k8scluster, ("_admin", "init"))
+            # helm-chart
+            if not init_target or "helm-chart" in init_target:
+                k8s_hc_id = None
+                try:
+                    k8s_hc_id, uninstall_sw = await self.helm_k8scluster.init_env(k8s_credentials,
+                                                                                  reuse_cluster_uuid=k8scluster_id)
+                    db_k8scluster_update["_admin.helm-chart.id"] = k8s_hc_id
+                    db_k8scluster_update["_admin.helm-chart.created"] = uninstall_sw
+                except Exception as e:
+                    error_text_list.append("Failing init helm-chart: {}".format(e))
+                    db_k8scluster_update["_admin.helm-chart.error_msg"] = str(e)
+                    if isinstance(e, K8sException):
+                        self.logger.error(logging_text + "Failing init helm-chart: {}".format(e))
+                    else:
+                        self.logger.error(logging_text + "Failing init helm-chart: {}".format(e), exc_info=True)
+
+            if not init_target or "juju-bundle" in init_target:
+                # Juju/k8s cluster
+                k8s_jb_id = None
+                try:
+                    k8s_jb_id, uninstall_sw = await self.juju_k8scluster.init_env(k8s_credentials)
+                    db_k8scluster_update["_admin.juju-bundle.id"] = k8s_jb_id
+                    db_k8scluster_update["_admin.juju-bundle.created"] = uninstall_sw
+                except Exception as e:
+                    error_text_list.append("Failing init juju-bundle: {}".format(e))
+                    db_k8scluster_update["_admin.juju-bundle.error_msg"] = str(e)
+                    if isinstance(e, N2VCException):
+                        self.logger.error(logging_text + "Failing init juju-bundle: {}".format(e))
+                    else:
+                        self.logger.error(logging_text + "Failing init juju-bundle: {}".format(e), exc_info=True)
+
+            # mark as an error if both helm-chart and juju-bundle have been failed
+            if k8s_hc_id or k8s_jb_id:
+                self.logger.debug(logging_text + "successfully created")
+                db_k8scluster_update["_admin.operationalState"] = "ENABLED"
+            else:
+                self.logger.debug(logging_text + "created with errors")
+                db_k8scluster_update["_admin.operationalState"] = "ERROR"
+                db_k8scluster_update["_admin.detailed-status"] = ";".join(error_text_list)
+
+        except Exception as e:
+            self.logger.critical(logging_text + "Exit Exception {}".format(e), exc_info=True)
+            exc = e
+        finally:
+            if exc and db_k8scluster:
+                db_k8scluster_update["_admin.operationalState"] = "ERROR"
+                db_k8scluster_update["_admin.detailed-status"] = "ERROR {}: {}".format(step, exc)
+
+                # Mark the k8scluster 'create' HA task as erroneous
+                operationState_HA = 'FAILED'
+                detailed_status_HA = "ERROR {}: {}".format(step, exc)
+            try:
+                if db_k8scluster_update:
+                    self.update_db_2("k8sclusters", k8scluster_id, db_k8scluster_update)
+
+                # Register the K8scluster 'create' HA task either
+                # succesful or erroneous, or do nothing (if legacy NBI)
+                self.lcm_tasks.register_HA('k8scluster', 'create', op_id,
+                                           operationState=operationState_HA,
+                                           detailed_status=detailed_status_HA)
+            except DbException as e:
+                self.logger.error(logging_text + "Cannot update database: {}".format(e))
+            self.lcm_tasks.remove("k8scluster", k8scluster_id, order_id)
+
+    async def delete(self, k8scluster_content, order_id):
+
+        # HA tasks and backward compatibility:
+        # If 'vim_content' does not include 'op_id', we a running a legacy NBI version.
+        # In such a case, HA is not supported by NBI, 'op_id' is None, and lock_HA() will do nothing.
+        # Register 'delete' task here for related future HA operations
+        op_id = k8scluster_content.pop('op_id', None)
+        if not self.lcm_tasks.lock_HA('k8scluster', 'delete', op_id):
+            return
+
+        k8scluster_id = k8scluster_content["_id"]
+        logging_text = "Task k8scluster_delete={} ".format(k8scluster_id)
+        self.logger.debug(logging_text + "Enter")
+
+        db_k8scluster = None
+        db_k8scluster_update = {}
+        exc = None
+        operationState_HA = ''
+        detailed_status_HA = ''
+        try:
+            step = "Getting k8scluster='{}' from db".format(k8scluster_id)
+            self.logger.debug(logging_text + step)
+            db_k8scluster = self.db.get_one("k8sclusters", {"_id": k8scluster_id})
+            k8s_hc_id = deep_get(db_k8scluster, ("_admin", "helm-chart", "id"))
+            k8s_jb_id = deep_get(db_k8scluster, ("_admin", "juju-bundle", "id"))
+
+            uninstall_sw = deep_get(db_k8scluster, ("_admin", "helm-chart", "created"))
+            cluster_removed = True
+            if k8s_jb_id:  # delete in reverse order of creation
+                step = "Removing juju-bundle '{}'".format(k8s_jb_id)
+                uninstall_sw = uninstall_sw or False
+                cluster_removed = await self.juju_k8scluster.reset(cluster_uuid=k8s_jb_id, uninstall_sw=uninstall_sw)
+                db_k8scluster_update["_admin.juju-bundle.id"] = None
+
+            if k8s_hc_id:
+                step = "Removing helm-chart '{}'".format(k8s_hc_id)
+                uninstall_sw = uninstall_sw or False
+                cluster_removed = await self.helm_k8scluster.reset(cluster_uuid=k8s_hc_id, uninstall_sw=uninstall_sw)
+                db_k8scluster_update["_admin.helm-chart.id"] = None
+
+            # Try to remove from cluster_inserted to clean old versions
+            if k8s_hc_id and cluster_removed:
+                step = "Removing k8scluster='{}' from k8srepos".format(k8scluster_id)
+                self.logger.debug(logging_text + step)
+                db_k8srepo_list = self.db.get_list("k8srepos", {"_admin.cluster-inserted": k8s_hc_id})
+                for k8srepo in db_k8srepo_list:
+                    try:
+                        cluster_list = k8srepo["_admin"]["cluster-inserted"]
+                        cluster_list.remove(k8s_hc_id)
+                        self.update_db_2("k8srepos", k8srepo["_id"], {"_admin.cluster-inserted": cluster_list})
+                    except Exception as e:
+                        self.logger.error("{}: {}".format(step, e))
+            self.db.del_one("k8sclusters", {"_id": k8scluster_id})
+            db_k8scluster_update = None
+            self.logger.debug(logging_text + "Done")
+
+        except Exception as e:
+            if isinstance(e, (LcmException, DbException, K8sException, N2VCException)):
+                self.logger.error(logging_text + "Exit Exception {}".format(e))
+            else:
+                self.logger.critical(logging_text + "Exit Exception {}".format(e), exc_info=True)
+            exc = e
+        finally:
+            if exc and db_k8scluster:
+                db_k8scluster_update["_admin.operationalState"] = "ERROR"
+                db_k8scluster_update["_admin.detailed-status"] = "ERROR {}: {}".format(step, exc)
+                # Mark the WIM 'create' HA task as erroneous
+                operationState_HA = 'FAILED'
+                detailed_status_HA = "ERROR {}: {}".format(step, exc)
+            try:
+                if db_k8scluster_update:
+                    self.update_db_2("k8sclusters", k8scluster_id, db_k8scluster_update)
+                # Register the K8scluster 'delete' HA task either
+                # succesful or erroneous, or do nothing (if legacy NBI)
+                self.lcm_tasks.register_HA('k8scluster', 'delete', op_id,
+                                           operationState=operationState_HA,
+                                           detailed_status=detailed_status_HA)
+            except DbException as e:
+                self.logger.error(logging_text + "Cannot update database: {}".format(e))
+            self.lcm_tasks.remove("k8scluster", k8scluster_id, order_id)
+
+
+class K8sRepoLcm(LcmBase):
+
+    def __init__(self, db, msg, fs, lcm_tasks, config, loop):
+        """
+        Init, Connect to database, filesystem storage, and messaging
+        :param config: two level dictionary with configuration. Top level should contain 'database', 'storage',
+        :return: None
+        """
+
+        self.logger = logging.getLogger('lcm.k8srepo')
+        self.loop = loop
+        self.lcm_tasks = lcm_tasks
+        self.vca_config = config["VCA"]
+        self.fs = fs
+        self.db = db
+
+        self.k8srepo = K8sHelmConnector(
+            kubectl_command=self.vca_config.get("kubectlpath"),
+            helm_command=self.vca_config.get("helmpath"),
+            fs=self.fs,
+            log=self.logger,
+            db=self.db,
+            on_update_db=None
+        )
+
+        super().__init__(db, msg, fs, self.logger)
+
+    async def create(self, k8srepo_content, order_id):
+
+        # HA tasks and backward compatibility:
+        # If 'vim_content' does not include 'op_id', we a running a legacy NBI version.
+        # In such a case, HA is not supported by NBI, 'op_id' is None, and lock_HA() will do nothing.
+        # Register 'create' task here for related future HA operations
+
+        op_id = k8srepo_content.pop('op_id', None)
+        if not self.lcm_tasks.lock_HA('k8srepo', 'create', op_id):
+            return
+
+        k8srepo_id = k8srepo_content.get("_id")
+        logging_text = "Task k8srepo_create={} ".format(k8srepo_id)
+        self.logger.debug(logging_text + "Enter")
+
+        db_k8srepo = None
+        db_k8srepo_update = {}
+        exc = None
+        operationState_HA = ''
+        detailed_status_HA = ''
+        try:
+            step = "Getting k8srepo-id='{}' from db".format(k8srepo_id)
+            self.logger.debug(logging_text + step)
+            db_k8srepo = self.db.get_one("k8srepos", {"_id": k8srepo_id})
+            db_k8srepo_update["_admin.operationalState"] = "ENABLED"
+        except Exception as e:
+            self.logger.critical(logging_text + "Exit Exception {}".format(e), exc_info=True)
+            exc = e
+        finally:
+            if exc and db_k8srepo:
+                db_k8srepo_update["_admin.operationalState"] = "ERROR"
+                db_k8srepo_update["_admin.detailed-status"] = "ERROR {}: {}".format(step, exc)
+                # Mark the WIM 'create' HA task as erroneous
+                operationState_HA = 'FAILED'
+                detailed_status_HA = "ERROR {}: {}".format(step, exc)
+            try:
+                if db_k8srepo_update:
+                    self.update_db_2("k8srepos", k8srepo_id, db_k8srepo_update)
+                # Register the K8srepo 'create' HA task either
+                # succesful or erroneous, or do nothing (if legacy NBI)
+                self.lcm_tasks.register_HA('k8srepo', 'create', op_id,
+                                           operationState=operationState_HA,
+                                           detailed_status=detailed_status_HA)
+            except DbException as e:
+                self.logger.error(logging_text + "Cannot update database: {}".format(e))
+            self.lcm_tasks.remove("k8srepo", k8srepo_id, order_id)
+
+    async def delete(self, k8srepo_content, order_id):
+
+        # HA tasks and backward compatibility:
+        # If 'vim_content' does not include 'op_id', we a running a legacy NBI version.
+        # In such a case, HA is not supported by NBI, 'op_id' is None, and lock_HA() will do nothing.
+        # Register 'delete' task here for related future HA operations
+        op_id = k8srepo_content.pop('op_id', None)
+        if not self.lcm_tasks.lock_HA('k8srepo', 'delete', op_id):
+            return
+
+        k8srepo_id = k8srepo_content.get("_id")
+        logging_text = "Task k8srepo_delete={} ".format(k8srepo_id)
+        self.logger.debug(logging_text + "Enter")
+
+        db_k8srepo = None
+        db_k8srepo_update = {}
+
+        exc = None
+        operationState_HA = ''
+        detailed_status_HA = ''
+        try:
+            step = "Getting k8srepo-id='{}' from db".format(k8srepo_id)
+            self.logger.debug(logging_text + step)
+            db_k8srepo = self.db.get_one("k8srepos", {"_id": k8srepo_id})
+
+        except Exception as e:
+            self.logger.critical(logging_text + "Exit Exception {}".format(e), exc_info=True)
+            exc = e
+        finally:
+            if exc and db_k8srepo:
+                db_k8srepo_update["_admin.operationalState"] = "ERROR"
+                db_k8srepo_update["_admin.detailed-status"] = "ERROR {}: {}".format(step, exc)
+                # Mark the WIM 'create' HA task as erroneous
+                operationState_HA = 'FAILED'
+                detailed_status_HA = "ERROR {}: {}".format(step, exc)
+            try:
+                if db_k8srepo_update:
+                    self.update_db_2("k8srepos", k8srepo_id, db_k8srepo_update)
+                # Register the K8srepo 'delete' HA task either
+                # succesful or erroneous, or do nothing (if legacy NBI)
+                self.lcm_tasks.register_HA('k8srepo', 'delete', op_id,
+                                           operationState=operationState_HA,
+                                           detailed_status=detailed_status_HA)
+                self.db.del_one("k8srepos", {"_id": k8srepo_id})
+            except DbException as e:
+                self.logger.error(logging_text + "Cannot update database: {}".format(e))
+            self.lcm_tasks.remove("k8srepo", k8srepo_id, order_id)