Convert orphan files deletion task to async using celery task. Update celery config to make it possible

Dockerfile

@@ -2,12 +2,12 @@ FROM python:3.9.20
 
 RUN apt-get update && apt-get install -y gcc build-essential && rm -rf /var/lib/apt/lists/*
 
-ENV PYTHONUNBUFFERED 1
+ENV PYTHONUNBUFFERED=1
 
 RUN curl -sSL https://install.python-poetry.org | python3 - --version 1.8.3
 # Poetry location so future commands (below) work
 
-ENV PATH $PATH:/root/.local/bin
+ENV PATH=$PATH:/root/.local/bin
 # Want poetry to use system python of docker container
 RUN poetry config virtualenvs.create false
 RUN poetry config virtualenvs.in-project false

pyproject.toml

@@ -19,7 +19,7 @@ channels = "2.4"
 channels-redis = "3.2.0"
 django-extra-fields = "0.9"
 pillow = "10.3.0"
-celery = "4.2.1"
+celery = "4.4.7"
 gunicorn = "22.0.0"
 urllib3 = ">=1.21.1,<1.25"
 uvicorn = {version = "0.13.3", extras = ["standard"]}

src/apps/analytics/tasks.py

@@ -1,3 +1,4 @@
+import os
 import time
 import logging
 from celery_config import app
@@ -570,3 +571,76 @@ def reset_computed_storage_analytics():
             elapsed_time
         )
     )
+
+
+@app.task(queue="site-worker")
+def delete_orphan_files():
+    logger.info("Task delete_orphan_files started")
+
+    # Find most recent file
+    most_recent_log_file = get_most_recent_storage_inconsistency_log_file(logger)
+    if not most_recent_log_file:
+        logger.warning("No storage inconsistency log file found. Nothing will be removed")
+        raise Exception("No storage inconsistency log file found")
+
+    # Get the list of orphan files from the content of the most recent log file
+    log_folder = "/app/logs/"
+    orphan_files_path = get_files_path_from_orphan_log_file(os.path.join(log_folder, most_recent_log_file), logger)
+
+    # Delete those files in batch (max 1000 element at once)
+    batch_size = 1000
+    for i in range(0, len(orphan_files_path), batch_size):
+        batch = orphan_files_path[i:i + batch_size]
+        objects_formatted = [{'Key': path} for path in batch]
+        BundleStorage.bucket.delete_objects(Delete={'Objects': objects_formatted})
+
+    logger.info("Delete oprhan files finished")
+
+
+def get_most_recent_storage_inconsistency_log_file(logger):
+    log_folder = "/app/logs/"
+    try:
+        log_files = [f for f in os.listdir(log_folder) if os.path.isfile(os.path.join(log_folder, f))]
+    except FileNotFoundError:
+        logger.info(f"Folder '{log_folder}' does not exist.")
+        return None
+
+    most_recent_log_file = None
+    most_recent_datetime = None
+    datetime_format = "%Y%m%d-%H%M%S"
+    for file in log_files:
+        try:
+            basename = os.path.basename(file)
+            datetime_str = basename[len("db_storage_inconsistency_"):-len(".log")]
+            file_datetime = datetime.strptime(datetime_str, datetime_format)
+            if most_recent_datetime is None or file_datetime > most_recent_datetime:
+                most_recent_datetime = file_datetime
+                most_recent_log_file = file
+        except ValueError:
+            logger.warning(f"Filename '{file}' does not match the expected format and will be ignored.")
+
+    return most_recent_log_file
+
+
+def get_files_path_from_orphan_log_file(log_file_path, logger):
+    files_path = []
+
+    try:
+        with open(log_file_path) as log_file:
+            lines = log_file.readlines()
+            orphan_files_lines = []
+            for i, line in enumerate(lines):
+                if "Orphaned files" in line:
+                    orphan_files_lines = lines[i + 1:]
+                    break
+
+            for orphan_files_line in orphan_files_lines:
+                files_path.append(orphan_files_line.split(maxsplit=1)[0])
+    except FileNotFoundError:
+        logger.error(f"File '{log_file_path}' does not exist.")
+    except PermissionError:
+        logger.error(f"Permission denied for reading the file '{log_file_path}'.")
+    except IOError as e:
+        logger.error(f"An I/O error occurred while accessing the file at {log_file_path}: {e}")
+
+    return files_path

src/apps/api/urls.py

@@ -69,6 +69,7 @@
     path('analytics/users_usage/', analytics.users_usage, name='users_usage'),
     path('analytics/delete_orphan_files/', analytics.delete_orphan_files, name="delete_orphan_files"),
     path('analytics/get_orphan_files/', analytics.get_orphan_files, name="get_orphan_files"),
+    path('analytics/check_orphans_deletion_status/', analytics.check_orphans_deletion_status, name="check_orphans_deletion_status"),
 
     # API Docs
     re_path(r'docs(?P<format>\.json|\.yaml)$', schema_view.without_ui(cache_timeout=0), name='schema-json'),

src/apps/api/views/analytics.py

@@ -12,7 +12,7 @@
 from competitions.models import Competition, Submission
 from analytics.models import StorageUsageHistory, CompetitionStorageDataPoint, UserStorageDataPoint
 from api.serializers.analytics import AnalyticsSerializer
-from utils.storage import BundleStorage
+from apps.analytics.tasks import delete_orphan_files as delete_orphan_files_async_task
 
 import os
 import datetime
@@ -22,6 +22,7 @@
 
 
 User = get_user_model()
+delete_orphan_files_task = None
 
 
 class SimpleFilterBackend(BaseFilterBackend):
@@ -322,36 +323,16 @@ def get_orphan_files(request):
 @api_view(["DELETE"])
 def delete_orphan_files(request):
     """
-    Delete all orphan files from the storage based on the last storage analytics
+    Start the deletion of orphan files task
     """
 
     if not request.user.is_superuser:
         raise PermissionDenied(detail="Admin only")
 
-    logger = logging.getLogger(__name__)
-    logger.info("Delete orphan files started")
-
-    # The analytics task generates a db_storage_inconsistency_<date>-<time>.log file that lists, among other things, the orphan files. Let's use it
-
-    # Find most recent file
-    most_recent_log_file = get_most_recent_storage_inconsistency_log_file()
-    if not most_recent_log_file:
-        logger.warning("No storage inconsistency log file found. Nothing will be removed")
-        return Response({"message": "No storage inconsistency log file found. Nothing will be removed"}, status=status.HTTP_404_NOT_FOUND)
-
-    # Get the list of orphan files from the content of the most recent log file
-    log_folder = "/app/logs/"
-    orphan_files_path = get_files_path_from_orphan_log_file(os.path.join(log_folder, most_recent_log_file))
-
-    # Delete those files in batch (max 1000 element at once)
-    batch_size = 1000
-    for i in range(0, len(orphan_files_path), batch_size):
-        batch = orphan_files_path[i:i + batch_size]
-        objects_formatted = [{'Key': path} for path in batch]
-        BundleStorage.bucket.delete_objects(Delete={'Objects': objects_formatted})
+    global delete_orphan_files_task
+    delete_orphan_files_task = delete_orphan_files_async_task.delay()
 
-    logger.info("Delete oprhan files finished")
-    return Response({"message": "done"}, status=status.HTTP_200_OK)
+    return Response({"success": True, "message": "orphan files deletion started"}, status=status.HTTP_200_OK)
 
 
 def get_most_recent_storage_inconsistency_log_file():
@@ -405,3 +386,21 @@ def get_files_path_from_orphan_log_file(log_file_path):
         logger.error(f"An I/O error occurred while accessing the file at {log_file_path}: {e}")
 
     return files_path
+
+
+@api_view(["GET"])
+def check_orphans_deletion_status(request):
+    """
+    Get the orphan files deletion task status.
+    Return one of ["PENDING", "STARTED", "SUCCESS", "FAILURE", "RETRY", "REVOKED"]
+    """
+
+    if not request.user.is_superuser:
+        raise PermissionDenied(detail="Admin only")
+
+    global delete_orphan_files_task
+    state = None
+    if delete_orphan_files_task:
+        state = delete_orphan_files_task.state
+
+    return Response({"status": state}, status=status.HTTP_200_OK)

src/settings/base.py

@@ -211,6 +211,8 @@
 CELERY_BROKER_URL = os.environ.get('BROKER_URL')
 if not CELERY_BROKER_URL:
     CELERY_BROKER_URL = f'pyamqp://{RABBITMQ_DEFAULT_USER}:{RABBITMQ_DEFAULT_PASS}@{RABBITMQ_HOST}:{RABBITMQ_PORT}//'
+CELERY_RESULT_BACKEND = os.environ.get("REDIS_URL", "redis://redis:6379")
+CELERY_IGNORE_RESULT = False      # Ensure that Celery tracks the state
 CELERY_TASK_SERIALIZER = 'json'
 CELERY_ACCEPT_CONTENT = ('json',)
 CELERY_BEAT_SCHEDULE = {

src/static/js/ours/client.js

@@ -352,6 +352,9 @@ CODALAB.api = {
     get_orphan_files: () => {
         return CODALAB.api.request('GET', `${URLS.API}analytics/get_orphan_files/`)
     },
+    check_orphans_deletion_status: () => {
+        return CODALAB.api.request('GET', `${URLS.API}analytics/check_orphans_deletion_status/`)
+    },
     /*---------------------------------------------------------------------
          User Quota and Cleanup
     ---------------------------------------------------------------------*/

src/static/riot/analytics/analytics.tag

@@ -133,8 +133,9 @@
             <a class="item" data-tab="competitions-usage">Competitions usage</a>
             <a class="item" data-tab="users-usage">Users usage</a>
             <div class="delete-oprhans-container">
-                <button class="ui red button" onclick="{showConfirmationModal}">
-                    <i class="icon warning"></i>Delete orphan files
+                <button class="ui red button { disabled: delete_orphans_button_modal_disabled }" onclick="{showConfirmationModal}">
+                    <i class="icon { warning: !delete_orphans_button_modal_loading}"></i>
+                    {delete_orphans_button_modal_text}
                 </button>
             </div>
         </div>
@@ -163,7 +164,7 @@
             <h3>Do you want to proceed ?</h3>
         </div>
         <div class="actions">
-            <button class="ui icon button {delete_button_color} { loading: delete_button_loading } { disabled: delete_button_disabled }" onclick="{deleteOrphanFiles}">
+            <button class="ui icon button {delete_button_color} { disabled: delete_button_disabled }" onclick="{deleteOrphanFiles}">
                 <i if={delete_button_color=="green"} class="check icon"></i>
                 {delete_button_text}
             </button>
@@ -217,6 +218,11 @@
         self.delete_button_disabled = false
         self.delete_button_text = "Yes, delete all orphan files"
 
+        self.delete_orphans_button_modal_text = "Delete orphan files"
+        self.delete_orphans_button_modal_loading = false
+        self.delete_orphans_button_modal_disabled = false
+        self.pollingInterval;
+
         self.one("mount", function () {
             // Semantic UI
             $('.tabular.menu .item', self.root).tab();
@@ -327,6 +333,7 @@
             self.time_range_shortcut("month");
             self.update_chart_resolution("day");
             self.getOrphanFiles();
+            self.startCheckOrphansDeletionStatus();
         })
 
         /*---------------------------------------------------------------------
@@ -525,23 +532,103 @@
             self.update();
         }
 
+        self.checkOrphansDeletionStatus = function() {
+            CODALAB.api.check_orphans_deletion_status()
+                .done(function(data) {
+                    if (data.status) {
+                        if (data.status == "SUCCESS") {
+                            toastr.success("Orphan files deletion successful")
+                            self.delete_button_color = "green";
+                            self.delete_button_text = "Deletion Successful";
+                        }
+                        if (data.status == "FAILURE") {
+                            toastr.error("Orphan files deletion failed")
+                            self.delete_button_color = "red";
+                            self.delete_button_text = "Deletion Failed";
+                        }
+                        if (data.status == "REVOKED") {
+                            toastr.error("Orphan files deletion has been canceled")
+                            self.delete_button_color = "red";
+                            self.delete_button_text = "Deletion canceled";
+                        }
+                        if (data.status == "SUCCESS" || data.status == "FAILURE" || data.status == "REVOKED") {
+                            // Task is over
+                            self.stopCheckOrphansDeletionStatus();
+                            self.delete_orphans_button_modal_text = "Delete orphan files";
+                            self.delete_orphans_button_modal_loading = false;
+                            self.delete_orphans_button_modal_disabled = false;
+                            self.delete_button_loading = false;
+                            self.delete_button_disabled = true;
+                        } else {
+                            // Task is running
+                            self.delete_orphans_button_modal_text = "Orphan files deletion in progress...";
+                            self.delete_orphans_button_modal_disabled = true;
+                            self.delete_orphans_button_modal_loading = true;
+
+                            self.delete_button_text = "Orphan files deletion in progress...";
+                            self.delete_button_disabled = true;
+                            self.delete_button_loading = true;
+                        }
+                    } else {
+                        // No task running
+                        self.stopCheckOrphansDeletionStatus();
+
+                        self.delete_orphans_button_modal_text = "Delete orphan files";
+                        self.delete_orphans_button_modal_disabled = false;
+                        self.delete_orphans_button_modal_loading = false;
+
+                        self.delete_button_color = "red";
+                        self.delete_button_loading = false;
+                        self.delete_button_disabled = false;
+                        self.delete_button_text = "Yes, delete all orphan files";
+                    }
+
+                })
+                .fail(function(response) {
+                    toastr.error("Orphan files deletion's task status check failed")
+                    self.delete_orphans_button_modal_text = "Delete orphan files";
+                    self.delete_orphans_button_modal_loading = false;
+                    self.delete_orphans_button_modal_disabled = false;
+
+                    self.delete_button_text = "Yes, delete all orphan files";
+                    self.delete_button_color = "red";
+                    self.delete_button_loading = false;
+                    self.delete_button_disabled = false;
+
+                    self.stopCheckOrphansDeletionStatus();
+                })
+                .always(function() {
+                    self.update();
+                })
+        }
+
+        self.startCheckOrphansDeletionStatus = function () {
+            self.pollingInterval = setInterval(self.checkOrphansDeletionStatus, 2000);
+        }
+
+        self.stopCheckOrphansDeletionStatus = function() {
+            if (self.pollingInterval) {
+                clearInterval(self.pollingInterval);
+                self.pollingInterval = null;
+            }
+        }
+
         self.deleteOrphanFiles = function() {
             self.delete_button_loading = true
             self.delete_button_disabled = true
+            self.delete_orphans_button_modal_loading = true;
+            self.delete_orphans_button_modal_disabled = true;
+            self.delete_button_text = "Orphan files deletion in progress...";
+            self.delete_orphans_button_modal_text = "Orphan files deletion in progress...";
             self.update()
             CODALAB.api.delete_orphan_files()
                 .done(function (data) {
-                    console.log("done", data);
-                    self.delete_button_color = "green";
-                    self.delete_button_disabled = true;
-                    self.delete_button_text = "Deletion Successful";
+                    if (data && data.success && !self.pollingInterval) {
+                        self.startCheckOrphansDeletionStatus();
+                    }
                 })
                 .fail(function (response) {
-                    console.log("fail response=", response);
-                    toastr.error("Deletion failed, error occurred")
-                    self.delete_button_color = "red";
-                    self.delete_button_disabled = false;
-                    self.delete_button_text = "Deletion Failed";
+                    toastr.error("Orphan files deletion failed to start")
                 })
                 .always(function () {
                     self.delete_button_loading = false
@@ -552,12 +639,10 @@
         self.getOrphanFiles = function() {
             CODALAB.api.get_orphan_files()
                 .done(function (data) {
-                    console.log("get_orphan_files success. Response", data);
                     self.nb_orphan_files = data.data.length
                     self.update({nb_orphan_files: self.nb_orphan_files});
                 })
                 .fail(function (response) {
-                    console.log("get_orphan_files failed. Response=", response);
                     toastr.error("Get oprhan files failed, error occurred")
                 });
         }