Open glennbech opened 6 years ago
The current way of using the module does not allow for full automation from the user's side. The user can create a bucket with s3, but then needs to incorporate build/packaging of the lambda and then uploading to the bucket before continuing with the rest of the module creation. Ideally, the module should refer to a publicly available bucket which contains the packages lambda that was build/released by the maintaining team.
We currently have a telia-oss
S3 bucket where we can deploy Lambda artifacts for OSS projects, so that they are publicly available. See e.g. https://github.com/telia-oss/concourse-sts-lambda/blob/master/.travis.yml
Once the artifact is uploaded to said bucket, it will be replicated to region specific buckets so that it can be used in any region. E.g.: https://github.com/telia-oss/concourse-sts-lambda/blob/master/terraform/modules/lambda/main.tf#L9-L10
I can help you set this up @glennbech, if you plan on moving the two modules into a single repository?
The Cloud Watch log forwarder is split between two repositories. One repository for the Terraform component, and another for the lambda function written in Java.
It would be a good idea to version them together.