set up an Amazon machine that can act as any production machine

[emanuil-tolev]

• [x] create and run an Ubuntu 12.04 LTS 64-bit AMI
• [ ] artforecaster
• [ ] leaps
• [ ] swap
• [ ] oag
• [ ] cl website
• [ ] etherpads
• [ ] save the resulting AMI
• [ ] gain access to CL account as an authorised user under CL's IAM
• [ ] transfer AMI to CL account
• [ ] consider backups of data - to EBS directly (push)? via this machine (pull)? Push seems preferred and much lower cost since this machine won't run most of the time. Keep in mind backups need to be able to be restored as well as created, an issue with elasticsearch. Create a new sysadmin issue for this
• [ ] consider sync issues - with this AMI being dormant most of the time, people need to remember to git pull, get their data and so on. What amount of this can be automated? Can a script launch this for 1 hour every day and do all the pulling of code and data that's needed around 3AM?
• [ ] are we backing up data - some services are OK with only code, some are useless without the data

[emanuil-tolev]

Investigate / report (incl. pricing) later on the following Amazon services related to this task:

• [x] Route 53, DNS: about £25 per month with Elastic Load Balancing
• [x] Elastic Load Balancing, so we don't have to wait for DNS: about £25 per month with Route 53
• [ ] Data Pipeline, potentially for backups https://aws.amazon.com/datapipeline/
• [ ] CloudWatch http://aws.amazon.com/cloudwatch/ (potential use: set custom metric, use PUT API to tell CloudWatch that Mythic Beasts server and associated running services are OK, use CloudWatch to start the backup production machine, sync code and data)
• [x] EC2 dedicated instances - isolated server, if we want to comply with some data storage regs. No dice, too expensive (base 2$ per hour and then 0.5-2$ p/h depending on size of instance).