S3 credentials are configured correctly for Paperclip:
# production.rb (same in development.rb)
config.paperclip_defaults = {
:storage => :s3,
:path => 'photos/:id/:style/:filename',
:s3_credentials => {
:bucket => ENV['aws_bucket'],
:access_key_id => ENV['aws_access_key'],
:secret_access_key => ENV['aws_secret_key']
}
}
I do see the environment variables set in the Heroku UI, and I can upload photos just fine.
However when I try to destroy the given model on production, I get this error in the log:
2016-01-14T15:48:48.079923+00:00 app[web.1]: Completed 500 Internal Server Error in 591ms (ActiveRecord: 60.3ms)
2016-01-14T15:48:48.082519+00:00 app[web.1]:
2016-01-14T15:48:48.082522+00:00 app[web.1]: AWS::Errors::MissingCredentialsError (
2016-01-14T15:48:48.082523+00:00 app[web.1]: Missing Credentials.
2016-01-14T15:48:48.082524+00:00 app[web.1]:
2016-01-14T15:48:48.082525+00:00 app[web.1]: Unable to find AWS credentials. You can configure your AWS credentials
2016-01-14T15:48:48.082526+00:00 app[web.1]: a few different ways:
2016-01-14T15:48:48.082526+00:00 app[web.1]:
2016-01-14T15:48:48.082527+00:00 app[web.1]: * Call AWS.config with :access_key_id and :secret_access_key
2016-01-14T15:48:48.082528+00:00 app[web.1]:
2016-01-14T15:48:48.082529+00:00 app[web.1]: * Export AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY to ENV
2016-01-14T15:48:48.082529+00:00 app[web.1]:
2016-01-14T15:48:48.082530+00:00 app[web.1]: * On EC2 you can run instances with an IAM instance profile and credentials
2016-01-14T15:48:48.082531+00:00 app[web.1]: will be auto loaded from the instance metadata service on those
2016-01-14T15:48:48.082531+00:00 app[web.1]: instances.
2016-01-14T15:48:48.082532+00:00 app[web.1]:
2016-01-14T15:48:48.082533+00:00 app[web.1]: * Call AWS.config with :credential_provider. A credential provider should
2016-01-14T15:48:48.082533+00:00 app[web.1]: either include AWS::Core::CredentialProviders::Provider or respond to
2016-01-14T15:48:48.082534+00:00 app[web.1]: the same public methods.
This all works just fine locally too (obviously using a different S3 bucket). What's the issue? Why am I getting AWS::Errors::MissingCredentialsError
just on production even though uploads work fine?