doug

Maybe - you wrote it!! I suggested testing the whole script on the instance first, before porting it to the CloudFormation template. My guess is that the 'sudo bash -c' portion isn't needed since we're already in bash and I think we're root - not sure though, so you should test.

"UserData" : { "Fn::Base64" : { "Fn::Join" : [ "", [ "#!/bin/bash -ex\n", "exec > >(tee /var/log/user-data.log|logger -t user-data -s 2>/dev/console) 2>&1\n", "resize2fs /dev/xvda1\n", "sudo bash -c 'echo \"{\\\"queue_url\\\":\\\"", { "Ref" : "SQSQueueURL" } , "'" ]] }}

This file is written by the init rake task defined in the Rakefile.rb file. What's being written is returned by the init method in the service/init/aws.rb class. But in this case null is being returned and nothing written since the user data isn't parseable as JSON. In terms of your script, I would suggest testing it out on a running instance then porting it into the CloudFormation template, escaping as needed. I'm guessing you would also want to populate the queue_url from the CloudFormation parameters...

Well, that user data will obviously not parse as JSON, but this should not cause Movie Masher to fail during initialization. Your shell script would need to also write out the JSON file to /var/www/config/userdata.json containing the queue url and your overrides.

The EC2 instance terminated? Perhaps you had more than one running in the pool and the one you were on was terminated by the auto scaling?

You should feel free to experiment with other instance types - the defaults from the cloudformation templates are only a suggestion.

Only changes the dimensions would affect the intermediate files. The bit rate is only applied after the intermediate files are generated and combined back together.

Directories should get automatically created. Again, there are no options to control the intermediate formats. These formats were chosen because they work in all cases. Other fomats will probably not work in all cases, and would likely reduce final output quality due to double compression. The size of your input media would not actually come into play here. In all cases the dimensions of these concat files should match the output dimensions, not the input dimensions. That's why I suggested reducing...