Introduction
You are responsible for your organization's email and web servers. It is
Saturday night and you are ready to enjoy the evening. Then, your pager goes off--it's work.
You call and get the "my email is broke" line. Hoping it is something simple, you
attempt to login; however, there is no response. You drive to work and discover
the hard drive has failed on the email server and all data is lost. Your evening is now wrecked.
Instead of having a fun evening, you will be spending your night installing a new
hard drive, reinstalling the operating system, and hoping the back-up does not
take all night (you do have a backup, right?).
It sure would be nice to have a failover system. In the case that the main system
fails, a slave system would take over the duties while you fix the master. This
solution should be inexpensive yet reliable. The solution should provide un-interrupted
service to the users while giving you a chance to repair on a more friendly time.
But, where do we find such a robust system that is inexpensive? Moreover, will it
work long enough to allow the maintenance to be addressed during "normal business hours"?
This is where the Linux High-Availability Project enters.
The goal is to have two
inexpensive Linux servers in a master/slave configuration to provide redundancy and flexibility
with critical services (such as email and web servers). When the slave detects
a failure of the master, the slave assumes a shared IP address and then starts the appropriate
services. When the slave detects the master is properly back online, the slave relinquishes the
services and IP address to the master. The Linux-HA project provides the software to handle
the fail-over between the master/slave servers.
Example of what goes next
Blah blah blah.
Conclusion
The end.
|