Add mutex lock file to registry CLI wrapper class

stack_orchestrator/deploy/deployment_create.py

@@ -422,10 +422,9 @@ def _copy_files_to_directory(file_paths: List[Path], directory: Path):
         copy(path, os.path.join(directory, os.path.basename(path)))
 
 
-def _create_deployment_file(deployment_dir: Path, cluster):
+def _create_deployment_file(deployment_dir: Path):
     deployment_file_path = deployment_dir.joinpath(constants.deployment_file_name)
-    if cluster is None:
-        cluster = f"{constants.cluster_name_prefix}{token_hex(8)}"
+    cluster = f"{constants.cluster_name_prefix}{token_hex(8)}"
     with open(deployment_file_path, "w") as output_file:
         output_file.write(f"{constants.cluster_id_key}: {cluster}\n")
 
@@ -474,7 +473,7 @@ def create_operation(deployment_command_context, spec_file, deployment_dir, netw
     # Copy spec file and the stack file into the deployment dir
     copyfile(spec_file, deployment_dir_path.joinpath(constants.spec_file_name))
     copyfile(stack_file, deployment_dir_path.joinpath(constants.stack_file_name))
-    _create_deployment_file(deployment_dir_path, deployment_command_context.cluster_context.cluster)
+    _create_deployment_file(deployment_dir_path)
     # Copy any config varibles from the spec file into an env file suitable for compose
     _write_config_file(spec_file, deployment_dir_path.joinpath(constants.config_file_name))
     # Copy any k8s config file into the deployment dir

stack_orchestrator/deploy/webapp/registry_mutex.py

@@ -1,59 +1,8 @@
+import fcntl
 from functools import wraps
-import os
-import time
 
 # Define default file path for the lock
 DEFAULT_LOCK_FILE_PATH = "/tmp/registry_mutex_lock_file"
-LOCK_TIMEOUT = 30
-LOCK_RETRY_INTERVAL = 3
-
-
-def acquire_lock(client, lock_file_path, timeout):
-    # Lock alreay acquired by the current client
-    if client.mutex_lock_acquired:
-        return
-
-    while True:
-        try:
-            # Check if lock file exists and is potentially stale
-            if os.path.exists(lock_file_path):
-                with open(lock_file_path, 'r') as lock_file:
-                    timestamp = float(lock_file.read().strip())
-
-                # If lock is stale, remove the lock file
-                if time.time() - timestamp > timeout:
-                    print(f"Stale lock detected, removing lock file {lock_file_path}")
-                    os.remove(lock_file_path)
-                else:
-                    print(f"Lock file {lock_file_path} exists and is recent, waiting...")
-                    time.sleep(LOCK_RETRY_INTERVAL)
-                    continue
-
-            # Try to create a new lock file with the current timestamp
-            fd = os.open(lock_file_path, os.O_CREAT | os.O_EXCL | os.O_RDWR)
-            with os.fdopen(fd, 'w') as lock_file:
-                lock_file.write(str(time.time()))
-
-            client.mutex_lock_acquired = True
-            print(f"Registry lock acquired, {lock_file_path}")
-
-            # Lock successfully acquired
-            return
-
-        except FileExistsError:
-            print(f"Lock file {lock_file_path} exists, waiting...")
-            time.sleep(LOCK_RETRY_INTERVAL)
-
-
-def release_lock(client, lock_file_path):
-    try:
-        os.remove(lock_file_path)
-
-        client.mutex_lock_acquired = False
-        print(f"Registry lock released, {lock_file_path}")
-    except FileNotFoundError:
-        # Lock file already removed
-        pass
 
 
 def registry_mutex():
@@ -64,13 +13,18 @@ def registry_mutex():
             if self.mutex_lock_file:
                 lock_file_path = self.mutex_lock_file
 
-            # Acquire the lock before running the function
-            acquire_lock(self, lock_file_path, LOCK_TIMEOUT)
-            try:
-                return func(self, *args, **kwargs)
-            finally:
-                # Release the lock after the function completes
-                release_lock(self, lock_file_path)
+            with open(lock_file_path, 'w') as lock_file:
+                try:
+                    # Try to acquire the lock
+                    fcntl.flock(lock_file, fcntl.LOCK_EX)
+
+                    # Call the actual function
+                    result = func(self, *args, **kwargs)
+                finally:
+                    # Always release the lock
+                    fcntl.flock(lock_file, fcntl.LOCK_UN)
+
+            return result
 
         return wrapper
 

stack_orchestrator/deploy/webapp/util.py

@@ -117,6 +117,7 @@ class LaconicRegistryClient:
     def __init__(self, config_file, log_file=None, mutex_lock_file=None):
         self.config_file = config_file
         self.log_file = log_file
+        self.mutex_lock_file = mutex_lock_file
         self.cache = AttrDict(
             {
                 "name_or_id": {},
@@ -125,9 +126,6 @@ class LaconicRegistryClient:
             }
         )
 
-        self.mutex_lock_file = mutex_lock_file
-        self.mutex_lock_acquired = False
-
     def whoami(self, refresh=False):
         if not refresh and "whoami" in self.cache:
             return self.cache["whoami"]
@@ -846,21 +844,16 @@ def confirm_payment(laconic: LaconicRegistryClient, record, payment_address, min
         )
         return False
 
-    # Check if the payment was already used on a deployment
+    # Check if the payment was already used on a
     used = laconic.app_deployments(
-        {"deployer": record.attributes.deployer, "payment": tx.hash}, all=True
+        {"deployer": payment_address, "payment": tx.hash}, all=True
     )
     if len(used):
-        # Fetch the app name from request record
-        used_request = laconic.get_record(used[0].attributes.request, require=True)
-
-        # Check that payment was used for deployment of same application
-        if record.attributes.application != used_request.attributes.application:
-            logger.log(f"{record.id}: payment {tx.hash} already used on a different application deployment {used}")
-            return False
+        logger.log(f"{record.id}: payment {tx.hash} already used on deployment {used}")
+        return False
 
     used = laconic.app_deployment_removals(
-        {"deployer": record.attributes.deployer, "payment": tx.hash}, all=True
+        {"deployer": payment_address, "payment": tx.hash}, all=True
     )
     if len(used):
         logger.log(