Background
It's been long since we have wondered why Amazon AWS infrastructure didn't provide for HA based ip failover. The real issue is due to the fact that 'CTDB' by itself doesn't provide any specific tools for these purposes. As we know that the overall management with in any given AWS `instance` happens through 'ec2-tools', CTDB also needs in turn access to these tools to provide IP failover.
So i worked on adding some functionality and here is the link to that project CTDB VIP support
This project is about collecting scripts to support VIP (Virtual IP) and CTDB based HA on AWS infrastructure. Currently tested and supported configuration is using the instances running under VPC (Virtual Private Cloud).
Usage
To start with configure ec2-tools on any given instance - Amazon docs have extensive documentation about SettingUp EC2 CommandLine.
After setting up EC2 tools, you should be able to run any ec2-tools without errors, if they still report errors - you should go back to Amazon documentation for further inputs.
On your Linux AWS instance
$ wget http://goo.gl/zGLpqa -O ctdb-ec2-1.1.tar.gz
$ rpmbuild -ta ctdb-ec2-1.1.tar.gz
$ yum localinstall ~/rpmbuild/RPMS/noarch/ctdb-ec2*noarch.rpm
Configuration
Once you have installed 'ctdb-ec2' package you have to add now your AWS credentials for everything to be working properly.
$ cat /etc/ctdb/ec2-config
## EDIT THIS FILE
export AWS_ACCESS_KEY=<EDIT>
export AWS_SECRET_KEY=<EDIT>
export JAVA_HOME=<java_installation_directory>
export EC2_HOME=<ec2_tools_installation_directory>
export EC2_URL=<ec2_region_url> ## example --> "https://ec2.us-west-1.amazonaws.com"
export PATH=$PATH:$EC2_HOME/binRepeat the steps listed in 'Usage and Configuration' on all the nodes which are part of the CTDB cluster.
Lastly restart 'ctdb' service after completing 'CTDB' configuration.
$ chkconfig ctdb on
$ service ctdb restart
