amazon

Today I was trying to setup a site behind an Elastic Load Balancer (ELB) on Amazon's EC2. I was able to set the configuration array to on in settings.php:

$conf['reverse_proxy'] = TRUE;

However, ELB's within AWS are setup to auto scale, which in turn can make their IP addresses change.

Recently an EC2 instance we were using in a production environment was scheduled to go down for maintenance. Normally this would be fine and we would let AWS just handle this. However, their scheduled maintenance window was set for 21:00 UTC, which equates to 17:00 EST. In other words, right at the end of the work day. Not cool! In order to use a maintenance schedule of our choosing, we had to migrate our production instance into a different availability zone prior to Amazon's schedule. So, we setup a test instances and tested migrating.

While using Amazon's AWS system, we needed to give access to one S3 bucket. Todo so, we ended up using Amazon's IAM.

Some thoughts for setting up a high performance Drupal stack:

Needs:

• High availability for an estimated 4,800 users.
• Traffic will be modest with heavy spikes during certain times.
• Traffic will be mainly authenticated users.
• Needs to be 100% reliable, especially during the spikes.

Solutions

• One server dedicated to MySQL.
  • EC2 instance running MySQL and not much else?

In a related note to the backup restore that I performed yesterday, I decided to run the upgrade.sh script provided by community.aegir.com. First, I downloaded the script to my restored EC2 instance wget -O upgrade.sh 'http://drupalcode.org/project/provision.git/blob_plain/refs/tags/6.x-1.6:/upgrade.sh.txt' Next I logged in as the Aegir user and ran the uprade.sh script per the instructions: su -s /bin/sh aegir -c "sh upgrade.sh" The status message displayed ment

Today I had the unfortunate need to get a restored database from an Amazon EBS snapshot. To do this I logged into the Amazon AWS management console and determined which snapshot I needed to restore from. Once I highlighted the correct snapshot, I clicked the 'Create Volume' button. Next, I chose which availability zone the snapshot should reside in and click 'Create' Next, I spun up a new instance of the production server, using the same AMI (ami-294aa340).