Cannot find support for Task Definition predefined in AWS ECS Edited

[Ritesh Mukim]

Currently, the pipe wants us to write full description of the task in the JSON file, which makes us specify the environment variables also. So, if the support is provided to just add the Task Definition name as the parameter to the pipe which is already created in AWS ECS

[ktomk]

To which description field are you referring to to? Could you provide more reference to Atlassian Bitbucket Piplines YAML file documentation and/or elaborate a bit more/provide reference to Amazone Web Services  in specific which Amazon Web Service in particular does provide Pipe Parameters to Atlassian Bitbucket Cloud Pipelines Plugin Pipes?

[Ritesh Mukim]

I am talking about AWS Elastic Container Service (ECS).

Please refer to the link to the pipe - https://bitbucket.org/atlassian/aws-ecs-deploy/src/1.1.4/

It requires us to define the pipeline with a parameter named - TASK_DEFINITION.
Which requires a string value, i.e. the path to the JSON file containing the task definition in full. In the task definition, we will have to provide the environment variables too. But I have already configured a TASK in AWS, so there should be an option to deploy the same task again.

[ktomk]

Fair point for ECS to re-use the task definition in AWS already.

However how should the pipe know this? Don't get me wrong I'm a big fan of having these things straight forward, but unless (and it does not seem the case here by your feedback) the Pipe does not read that task definition from AWS ECS when it runs.

So this would be a feature request for the pipe. But I must admit I don't have any experience how (well) this works, I'd assume it will take some time.

As you can combine a Pipe with more Step script commands (before appears useful to me here, after is possible in general as well but not applicable here) maybe there is an option to make use of the AWS CLI utility (or other AWS API) in the pipeline step to fetch the existing task definition (store it as a file) and then point the Pipe to that file.

Or is that not such a good idea for your use-case? Would be a work-around for your use-case but also a good example for a feature request which then should be more easy to get done.