fix: added lock to prevent same task run multiple time at the same time (#784)

* fix: added lock to prevent same task run multiple time at the same time

* fix: add debounce strategy before calling the celery tasks to avoid queuing extra tasks

* chore: version bumped

Author: marslanabdulrauf

src/ol_openedx_git_auto_export/ol_openedx_git_auto_export/constants.py

@@ -5,3 +5,12 @@
 
 COURSE_RERUN_STATE_SUCCEEDED = "succeeded"
 REPOSITORY_NAME_MAX_LENGTH = 100  # Max length from GitHub for repo name
+
+# Debounce settings for the signal handler.
+# A single course save triggers 10-30 COURSE_PUBLISHED signals in one request.
+# cache.add() on this key ensures only the first signal schedules a task; all
+# subsequent signals within the window are silently dropped before hitting the broker.
+# The task is scheduled with countdown=EXPORT_DEBOUNCE_DELAY so it runs after
+# the burst window has closed and the course state is fully settled.
+EXPORT_DEBOUNCE_DELAY = 5  # seconds — must exceed the publish burst window
+EXPORT_DEBOUNCE_CACHE_KEY = "git_export_debounce:{course_key}"

src/ol_openedx_git_auto_export/ol_openedx_git_auto_export/tasks.py

@@ -16,6 +16,7 @@
 
 from ol_openedx_git_auto_export.models import CourseGitRepository
 from ol_openedx_git_auto_export.utils import (
+    clear_stale_git_lock,
     export_course_to_git,
     github_repo_name_format,
     is_auto_repo_creation_enabled,
@@ -39,6 +40,10 @@ def async_export_to_git(course_key_string, user=None):
                 "Starting async course content export to git (course id: %s)",
                 course_module.id,
             )
+            # Remove any stale .git/index.lock left by a previously crashed worker.
+            # Dirty working-tree files from a prior crash are cleaned by the
+            # `git reset --hard origin/<branch>` + `git clean` inside export_to_git.
+            clear_stale_git_lock(course_repo.git_url)
             export_to_git(course_module.id, course_repo.git_url, user=user)
         else:
             LOGGER.info(

src/ol_openedx_git_auto_export/ol_openedx_git_auto_export/utils.py

@@ -5,15 +5,19 @@
 import logging
 import os
 import re
+from pathlib import Path
 
 from django.conf import settings
 from django.contrib.auth.models import User
+from django.core.cache import cache
 from django.core.exceptions import ImproperlyConfigured
 from xmodule.modulestore.django import modulestore
 
 from ol_openedx_git_auto_export.constants import (
     ENABLE_AUTO_GITHUB_REPO_CREATION,
     ENABLE_GIT_AUTO_EXPORT,
+    EXPORT_DEBOUNCE_CACHE_KEY,
+    EXPORT_DEBOUNCE_DELAY,
     REPOSITORY_NAME_MAX_LENGTH,
 )
 
@@ -100,7 +104,41 @@ def export_course_to_git(course_key):
         )
 
         user = get_publisher_username(course_module)
-        async_export_to_git.delay(str(course_key), user)
+
+        debounce_key = EXPORT_DEBOUNCE_CACHE_KEY.format(course_key=str(course_key))
+        if cache.add(debounce_key, "1", timeout=EXPORT_DEBOUNCE_DELAY):
+            log.info(
+                "Scheduling git export for course %s with %ds debounce delay",
+                course_key,
+                EXPORT_DEBOUNCE_DELAY,
+            )
+            async_export_to_git.apply_async(
+                args=[str(course_key), user],
+                countdown=EXPORT_DEBOUNCE_DELAY,
+            )
+        else:
+            log.info(
+                "Git export already scheduled for course %s, skipping duplicate signal",
+                course_key,
+            )
+
+
+def clear_stale_git_lock(git_url):
+    """
+    Remove a stale .git/index.lock file for the local clone of git_url, if present.
+
+    A stale lock file can be left behind when a worker process is killed mid-operation.
+    """
+    git_repo_export_dir = getattr(
+        settings, "GIT_REPO_EXPORT_DIR", "/openedx/export_course_repos"
+    )
+    rdir = git_url.rsplit("/", 1)[-1].rsplit(".git", 1)[0]
+    index_lock = Path(git_repo_export_dir) / rdir / ".git" / "index.lock"
+    if index_lock.exists():
+        log.warning(
+            "Removing stale .git/index.lock for repo %s at %s", git_url, index_lock
+        )
+        index_lock.unlink()
 
 
 def is_auto_repo_creation_enabled():

src/ol_openedx_git_auto_export/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "ol-openedx-git-auto-export"
-version = "0.7.1"
+version = "0.7.2"
 description = "A plugin that auto saves the course OLX to git when an author publishes it"
 authors = [
   {name = "MIT Office of Digital Learning"}