def finish_reprocessing(project_id, group_id, new_group_id):
from sentry.models import Group, GroupRedirect, Activity
with transaction.atomic():
group = Group.objects.get(id=group_id)
new_group = Group.objects.get(id=new_group_id)
# Any sort of success message will be shown at the *new* group ID's URL
GroupRedirect.objects.create(
organization_id=new_group.project.organization_id,
group_id=new_group.id,
previous_group_id=group_id,
)
# While we migrated all associated models at the beginning of
# reprocessing, there is still the "reprocessing" activity that we need
# to transfer manually.
Activity.objects.filter(group_id=group_id).update(
group_id=new_group_id)
# All the associated models (groupassignee and eventattachments) should
# have moved to a successor group that may be deleted independently.
group.delete()
# Need to delay this until we have enqueued all events.
eventstream.exclude_groups(project_id, [group_id])
from sentry import similarity
similarity.delete(None, group)
def reprocess_group(
project_id, group_id, query_state=None, start_time=None, max_events=None, acting_user_id=None
):
from sentry.reprocessing2 import start_group_reprocessing
if start_time is None:
start_time = time.time()
start_group_reprocessing(
project_id, group_id, max_events=max_events, acting_user_id=acting_user_id
)
query_state, events = celery_run_batch_query(
filter=eventstore.Filter(project_ids=[project_id], group_ids=[group_id]),
batch_size=GROUP_REPROCESSING_CHUNK_SIZE,
state=query_state,
referrer="reprocessing2.reprocess_group",
)
if not events:
# Need to delay this until we have queried all events.
eventstream.exclude_groups(project_id, [group_id])
wait_group_reprocessed.delay(project_id=project_id, group_id=group_id)
return
tombstoned_event_ids = []
for event in events:
if max_events is None or max_events > 0:
reprocess_event.delay(
project_id=project_id, event_id=event.event_id, start_time=start_time,
)
if max_events is not None:
max_events -= 1
else:
tombstoned_event_ids.append(event.event_id)
# len(tombstoned_event_ids) is upper-bounded by GROUP_REPROCESSING_CHUNK_SIZE
if tombstoned_event_ids:
tombstone_events.delay(
project_id=project_id, group_id=group_id, event_ids=tombstoned_event_ids
)
reprocess_group.delay(
project_id=project_id,
group_id=group_id,
query_state=query_state,
start_time=start_time,
max_events=max_events,
)
def finish_reprocessing(project_id, group_id):
from sentry.models import Activity, Group, GroupRedirect
with transaction.atomic():
group = Group.objects.get(id=group_id)
# While we migrated all associated models at the beginning of
# reprocessing, there is still the "reprocessing" activity that we need
# to transfer manually.
activity = Activity.objects.get(group_id=group_id)
new_group_id = activity.group_id = activity.data["newGroupId"]
activity.save()
new_group = Group.objects.get(id=new_group_id)
# Any sort of success message will be shown at the *new* group ID's URL
GroupRedirect.objects.create(
organization_id=new_group.project.organization_id,
group_id=new_group_id,
previous_group_id=group_id,
)
# All the associated models (groupassignee and eventattachments) should
# have moved to a successor group that may be deleted independently.
group.delete()
# Tombstone unwanted events that should be dropped after new group
# is generated after reprocessing
buffered_delete_old_primary_hash(
project_id=project_id,
group_id=group_id,
force_flush_batch=True,
)
eventstream.exclude_groups(project_id, [group_id])
from sentry import similarity
similarity.delete(None, group)
def reprocess_group(
project_id,
group_id,
remaining_events="delete",
new_group_id=None,
query_state=None,
start_time=None,
max_events=None,
acting_user_id=None,
):
sentry_sdk.set_tag("project", project_id)
from sentry.reprocessing2 import (
CannotReprocess,
logger,
mark_event_reprocessed,
reprocess_event,
start_group_reprocessing,
)
if start_time is None:
assert new_group_id is None
start_time = time.time()
new_group_id = start_group_reprocessing(
project_id,
group_id,
max_events=max_events,
acting_user_id=acting_user_id,
remaining_events=remaining_events,
)
assert new_group_id is not None
query_state, events = celery_run_batch_query(
filter=eventstore.Filter(project_ids=[project_id],
group_ids=[group_id]),
batch_size=GROUP_REPROCESSING_CHUNK_SIZE,
state=query_state,
referrer="reprocessing2.reprocess_group",
)
if not events:
# Need to delay this until we have enqueued all events and stopped
# iterating over the batch query, if we take care of this in
# finish_reprocessing it won't work, as for small max_events
# finish_reprocessing may execute sooner than the last reprocess_group
# iteration.
eventstream.exclude_groups(project_id, [group_id])
return
remaining_event_ids = []
remaining_events_min_datetime = None
remaining_events_max_datetime = None
for event in events:
if max_events is None or max_events > 0:
with sentry_sdk.start_span(op="reprocess_event"):
try:
reprocess_event(
project_id=project_id,
event_id=event.event_id,
start_time=start_time,
)
except CannotReprocess as e:
logger.error(f"reprocessing2.{e}")
except Exception:
sentry_sdk.capture_exception()
else:
if max_events is not None:
max_events -= 1
continue
# In case of errors while kicking off reprocessing, mark the event
# as reprocessed such that progressbar advances and the
# finish_reprocessing task is still correctly spawned.
mark_event_reprocessed(group_id=group_id, project_id=project_id)
# In case of errors while kicking off reprocessing or if max_events has
# been exceeded, do the default action.
if remaining_events_min_datetime is None or remaining_events_min_datetime > event.datetime:
remaining_events_min_datetime = event.datetime
if remaining_events_max_datetime is None or remaining_events_max_datetime < event.datetime:
remaining_events_max_datetime = event.datetime
remaining_event_ids.append(event.event_id)
# len(remaining_event_ids) is upper-bounded by GROUP_REPROCESSING_CHUNK_SIZE
if remaining_event_ids:
handle_remaining_events.delay(
project_id=project_id,
new_group_id=new_group_id,
event_ids=remaining_event_ids,
remaining_events=remaining_events,
from_timestamp=remaining_events_min_datetime,
to_timestamp=remaining_events_max_datetime,
)
reprocess_group.delay(
project_id=project_id,
group_id=group_id,
new_group_id=new_group_id,
query_state=query_state,
start_time=start_time,
max_events=max_events,
remaining_events=remaining_events,
)