openstack
/
senlin
Code Issues Proposed changes

Fix action failed status 

Change _mark_failed to commit update to the action status before marking
the depended actions' status as failed.

This fixes a problem where a retry on deadlock for the depended action
was rolling back the update to the dependent action status.

Closes-Bug: 1808846
Change-Id: Icc4d6bf7034929eeb67d60d55b538b5f87bbba17
This commit is contained in:
Duc Truong 2018-12-17 17:57:14 +00:00
parent 29ed02378f
commit 38e71876ee
1 changed files with 18 additions and 18 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

36
senlin/db/sqlalchemy/api.py
View File

@ -1238,31 +1238,31 @@ def action_mark_ready(context, action_id, timestamp):
@retry_on_deadlock
def _mark_failed(session, action_id, timestamp, reason=None):
def _mark_failed(action_id, timestamp, reason=None):
# mark myself as failed
query = session.query(models.Action).filter_by(id=action_id)
values = {
'owner': None,
'status': consts.ACTION_FAILED,
'status_reason': (six.text_type(reason) if reason else
'Action execution failed'),
'end_time': timestamp,
}
query.update(values, synchronize_session=False)
with session_for_write() as session:
query = session.query(models.Action).filter_by(id=action_id)
values = {
'owner': None,
'status': consts.ACTION_FAILED,
'status_reason': (six.text_type(reason) if reason else
'Action execution failed'),
'end_time': timestamp,
}
query.update(values, synchronize_session=False)
query = session.query(models.ActionDependency)
query = query.filter_by(depended=action_id)
dependents = [d.dependent for d in query.all()]
query.delete(synchronize_session=False)
query = session.query(models.ActionDependency)
query = query.filter_by(depended=action_id)
dependents = [d.dependent for d in query.all()]
query.delete(synchronize_session=False)
for d in dependents:
_mark_failed(session, d, timestamp)
_mark_failed(d, timestamp)
@retry_on_deadlock
def action_mark_failed(context, action_id, timestamp, reason=None):
with session_for_write() as session:
_mark_failed(session, action_id, timestamp, reason)
_mark_failed(action_id, timestamp, reason)
@retry_on_deadlock
@ -1608,7 +1608,7 @@ def gc_by_engine(engine_id):
_release_cluster_lock(session, cl, a.id, 1)
# mark action failed and release lock
_mark_failed(session, a.id, timestamp, reason="Engine failure")
_mark_failed(a.id, timestamp, reason="Engine failure")
# HealthRegistry