Description of problem:
When deploying via the ucli, there's no way to specify the stack timeout, so the default value (1 hour, configurable via stack_action_timeout in heat.conf) is always used.
This is likely to be too low for large deployments, and it's less than the undercloud token expiry (set to 4 hours by default), so we should make it configurable via the CLI.
The heatclient convention for this is -t/--timeout
So we need something like:
openstack overcloud deploy --timeout 240
Which would result in a 4 hour timout (heatclient and the heat API expects the timeout in minutes)
The API call stack_args need to have timeout_mins added to them with this value.
Patch midstream https://review.gerrithub.io/#/c/239943/
To clarify the workaround, if this isn't configurable via the CLI, you can set the global timeout in the heat config to a higher value:
- Edit /etc/heat/heat.conf on the undercloud to specify stack_action_timeout = 14400 (4 hours in seconds, matches the undercloud keystone token expiration)
- Restart the openstack-heat-engine process
This will give a global 4 hour timeout when not specifying any timeout via the CLI/API.
I'm moving this back to GA since we have two other fixes we have to merge today anyway. It's been breaking a lot of our QE test deployments, I believe.
the timeout is specified in CLI with: -t <min>
tested with seeting it to 1 minute - the deployment exited after 1 minute.
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.
For information on the advisory, and where to find the updated
files, follow the link below.
If the solution does not work for you, open a new bug report.