fix(scheduler): scheduled tasks does not persist on server restart

kolibri/core/analytics/utils.py

@@ -503,14 +503,18 @@ def schedule_ping(
     server=DEFAULT_SERVER_URL,
     checkrate=DEFAULT_PING_CHECKRATE,
     interval=DEFAULT_PING_INTERVAL,
+    job_id=None,
 ):
-    started = local_now()
-    scheduler.schedule(
-        started,
-        _ping,
-        interval=interval * 60,
-        repeat=None,
-        started=started,
-        server=server,
-        checkrate=checkrate,
-    )
+    # If pinging is not disabled by the environment
+    if not conf.OPTIONS["Deployment"]["DISABLE_PING"]:
+        started = local_now()
+        scheduler.schedule(
+            started,
+            _ping,
+            interval=interval * 60,
+            repeat=None,
+            started=started,
+            server=server,
+            checkrate=checkrate,
+            job_id=job_id,
+        )

kolibri/core/deviceadmin/utils.py

@@ -210,11 +210,13 @@ def perform_vacuum(database=db.DEFAULT_DB_ALIAS):
             logger.info("Sqlite database Vacuum finished.")
 
 
-def schedule_vacuum():
+def schedule_vacuum(job_id=None):
     current_dt = local_now()
     vacuum_time = current_dt.replace(hour=3, minute=0, second=0, microsecond=0)
     if vacuum_time < current_dt:
         # If it is past 3AM, change the day to tomorrow.
         vacuum_time = vacuum_time + timedelta(days=1)
     # Repeat indefinitely
-    scheduler.schedule(vacuum_time, perform_vacuum, repeat=None, interval=24 * 60 * 60)
+    scheduler.schedule(
+        vacuum_time, perform_vacuum, repeat=None, interval=24 * 60 * 60, job_id=job_id
+    )

kolibri/core/tasks/job.py

@@ -87,7 +87,12 @@ def __init__(self, func, *args, **kwargs):
             kwargs["cancellable"] = func.cancellable
             kwargs["extra_metadata"] = func.extra_metadata.copy()
             func = func.func
-        self.job_id = uuid.uuid4().hex
+
+        job_id = kwargs.pop("job_id", None)
+        if job_id is None:
+            job_id = uuid.uuid4().hex
+
+        self.job_id = job_id
         self.state = kwargs.pop("state", State.QUEUED)
         self.traceback = ""
         self.exception = None

kolibri/utils/server.py

@@ -18,7 +18,6 @@
 from .system import pid_exists
 from kolibri.core.content.utils import paths
 from kolibri.core.content.zip_wsgi import get_application
-from kolibri.core.deviceadmin.utils import schedule_vacuum
 from kolibri.core.tasks.main import initialize_workers
 from kolibri.core.tasks.main import scheduler
 from kolibri.utils import conf
@@ -67,6 +66,10 @@
 # Currently non-configurable until we know how to properly handle this
 LISTEN_ADDRESS = "0.0.0.0"
 
+# Constant job_id for scheduled jobs that we want to keep track of across server restarts
+SCH_PING_JOB_ID = 0
+SCH_VACUUM_JOB_ID = 1
+
 
 class NotRunning(Exception):
     """
@@ -86,22 +89,22 @@ def __init__(self, bus, port):
         self.workers = None
 
     def start(self):
-        # Initialize the iceqube scheduler to handle scheduled tasks
-        scheduler.clear_scheduler()
-
-        if not conf.OPTIONS["Deployment"]["DISABLE_PING"]:
-
-            # schedule the pingback job
+        # schedule the pingback job if not already scheduled
+        if SCH_PING_JOB_ID not in scheduler:
             from kolibri.core.analytics.utils import schedule_ping
 
-            schedule_ping()
+            schedule_ping(job_id=SCH_PING_JOB_ID)
+
+        # schedule the vacuum job if not already scheduled
+        if SCH_VACUUM_JOB_ID not in scheduler:
+            from kolibri.core.deviceadmin.utils import schedule_vacuum
 
-        # schedule the vacuum job
-        schedule_vacuum()
+            schedule_vacuum(job_id=SCH_VACUUM_JOB_ID)
 
         # Initialize the iceqube engine to handle queued tasks
         self.workers = initialize_workers()
 
+        # Initialize the iceqube scheduler to handle scheduled tasks
         scheduler.start_scheduler()
 
         # Register the Kolibri zeroconf service so it will be discoverable on the network