Convergence: Immediate concurrent update fails

[Anant Patil]

I ran following command:

heat stack-create -f tr41.yaml tr1; sleep 1; heat stack-update -f tr31.yaml tr1; sleep 1; heat stack-update -f tr1.yaml tr1

where tr41 is a template with 5 TestResource resources, tr31 is template with 3 TestResource resources and tr1 has one resource.

The log messages are something like:

[3f0eaac1-d84c-4de6-92da-5208e0a207e0:c71251eb-4a7c-46e2-a036-877075ece336:True] Waiting 3f0eaac1-d84c-4de6-92da-5208e0a207e0: Got (429, False), (428, False); still need (418, False) sync /opt/stack/heat/heat/engine/sync_point.py:138

Found that the graph is not properly formed.

[Anant Patil (ananta)]

The issue is with updating the stack when the graph is being formed. Immediate update will result in malformed graph as the previous create/update was not complete. The solution is to lock the stack till graph and sync-points are created in DB.