Description of problem:
Currently, from a TripleO standpoint if the stack create is stuck for a long time at a particular step and the user knows the deployment is going to fail, the only way to cancel the update is to restart heat engine or wait until update is failed after the 240 minute timeout. heat-cancel-update puts things in a really bad state ROLLBACK_FAILED. It would be good to have a cli command to instruct heat to just stop the update.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
This has been discussed upstream, at least for convergence, but nobody got around to implementing it yet:
Probably not going to be feasible until convergence architecture is enabled in the TripleO undercloud. Looks like an Queens/OSP 13 feature horizon.
As far as I know, for this to work with convergence we just need to add a --no-rollback option and add it to the ReST API. Internally, the engine's implementation already allows cancelling a stack without rolling it back (this happens e.g. when a sibling resource in the parent stack fails).
SO provided we can enable convergence in TripleO for Queens (the hard part), this should be relatively easy to implement.
*** This bug has been marked as a duplicate of bug 1480410 ***