[SATLUG] Create High Availability SAN from scratch

jaret jaret at aberlorn.com
Mon Jun 14 18:48:25 CDT 2010


Enrique Sanchez wrote:
> On Mon, Jun 14, 2010 at 4:48 PM, John Pappas <j at jvpappas.net> wrote:
>   
>> Complex question, but:
>>
>>     
>
> I would add that you need to add a mechanism to detect when the other
> host gets hung and how to take it down to prevent data corruption or
> prevent the surviving node from taking over, also known as STONITH,
> short for Shoot The Other Node In The HEAD. this also would work on a
> split brain scenario but the heartbeat link(s), preferably using
> redundant & alternate paths,  as suggested above, should take care of
> it.
>
>   
Good. I understand this.
> it is pretty darn easy to fall for the RSA cards, but on a power
> failure situation, the RSA card would also loose power, having the
> potential to lock down your surviving node and preventing it from
> taking over (it just happened to me).
>   
So, let me try and put two-and-two together. Node 1 went down and this 
node had your usb-powered RSA card in it. The power-failure corrupted 
data and so you couldn't get Node 2 to take over, which created the 
STONITH scenario. Ouch. That sucks.
> if I have had a saying in the solution, I would have gone for an
> server independent method but it did not happen.
>   
If I'm following you right in the above paragraph, then by "server 
independent" you mean using a network-cluster manager, like PaceMaker 
cluster?
>
> best regards,
> esv.
>
>   



More information about the SATLUG mailing list