[ale] Redundant File Servers

Stephan Uphoff ups at tree.com
Tue Sep 14 13:47:31 EDT 2004


Read up on split-brain syndrome.
This happens when both servers think that the other one is dead
and access a shared storage device.
This can destroy a filesystem really, really fast.
Depending on your heartbeat solution this can happen as easily as
a server being to busy/low on memory,.. to send a heartbeat signal.
With a bad heartbeat solution your redundant server setup can 
be less reliable then a single server.
A good solution is the "shoot them in the head" approach:
If server B thinks server A is dead -> B cuts the power to A.
Now B can be sure that A is dead ;-)


On Tue, 2004-09-14 at 08:50, Kevin O'Neill Stoll wrote:
> Hi all,
> 
> I have been doing some research about either:
> a failover pair of fileservers, possibly load-balanced
> across the pair
> or
> a high-availability load-balanced cluster of file servers
> 
> I just wanted to know if anyone was already working with a
> particular setup or product that they have had success
> with. I have been doing all the basic reading / research
> that I can find via google, tldp, and linux-ha.org.
> 
> I have already found OpenGFS, a Networked Raid 1 type
> setup, and supposedly RedHat's cluster management package
> has a solution available.
> 
> Any thoughts would be greatly appreciated.
> 
> =====
> Kevin Stoll
> http://kevinstoll.com/
> 
> OpenSource Software...FREE!
> Angering Bill Gates...priceless.
> ============================================================
> 
> 
> 
> _______________________________
> Do you Yahoo!?
> Declare Yourself - Register online to vote today!
> http://vote.yahoo.com
> _______________________________________________
> Ale mailing list
> Ale at ale.org
> http://www.ale.org/mailman/listinfo/ale
> 
> 



More information about the Ale mailing list