When launching stack groups, if one of the stack config's templates is invalid, meaningful errors are not bubble up.
I have to launch the specific template to get error that needs to be fixed.
Finding the specific template to launch to get the error back can be frustrating and time consuming with multiple templates.
Mitchell@MitchellSuzuki MINGW64
$ sceptre launch beta\\service
Do you want to launch 'beta\service' [y/N]: y
[2019-02-16 23:14:41] - beta\service\rabbitmq - Launching Stack
[2019-02-16 23:14:41] - beta\service\audit - Launching Stack
[2019-02-16 23:14:41] - beta\service\ecs-cluster - Launching Stack
[2019-02-16 23:14:42] - beta\service\audit - Stack is in the PENDING state
[2019-02-16 23:14:42] - beta\service\audit - Creating Stack
[2019-02-16 23:14:42] - beta\service\rabbitmq - Stack is in the CREATE_COMPLETE state
[2019-02-16 23:14:42] - beta\service\rabbitmq - Updating Stack
[2019-02-16 23:14:42] - beta\service\ecs-cluster - Stack is in the PENDING state
[2019-02-16 23:14:42] - beta\service\ecs-cluster - Creating Stack
[2019-02-16 23:14:43] - beta\service\rabbitmq - No updates to perform.
Stack with id beta-service-ecs-cluster does not exist
(venv)
Mitchell@MitchellSuzuki
$ sceptre launch beta\\service\\ecs-cluster
Do you want to launch 'beta\service\ecs-cluster' [y/N]: y
[2019-02-16 23:14:57] - beta\service\ecs-cluster - Launching Stack
[2019-02-16 23:14:58] - beta\service\ecs-cluster - Stack is in the PENDING state
[2019-02-16 23:14:58] - beta\service\ecs-cluster - Creating Stack
An error occurred (ValidationError) when calling the CreateStack operation: Parameters: [RabbitSecurityGroup] do not exist in the template
Better, more useful error reporting is definitely something that has been front of mind. I think we probably need to adjust the response structure from executor or actions such that they are able to capture the response and any errors e.g.
{
"response": [],
"errors": []
}
Worth having a discussion on this point because we will apply the solution across the whole project.
This is causing me a great deal of grief currently, as I'm working on a project with a large number of stacks. I've had to resort to adding print statements within StackActions to find out which Stack is actually causing the error.
We have the same issue. What type of discussion would help this issue forward?
Yep this would be good to incorporate. Currently trying to catch up with things.
Most helpful comment
Better, more useful error reporting is definitely something that has been front of mind. I think we probably need to adjust the response structure from
executororactionssuch that they are able to capture the response and any errors e.g.Worth having a discussion on this point because we will apply the solution across the whole project.