Results 1 to 4 of 4
Hi
I am having a problem in using 3 network card.
my application is usng the first network card.
however, if the 1st network fails, how do i make the ...
Enjoy an ad free experience by logging in. Not a member yet? Register.
- 04-04-2004 #1Just Joined!
- Join Date
- Apr 2004
- Posts
- 10
3 network cards
Hi
I am having a problem in using 3 network card.
my application is usng the first network card.
however, if the 1st network fails, how do i make the 2nd network card to work such that my application will not fail.
is there something called "teaming"?
How do i make the fail over for the network card such that if the 1st card fails, the 2nd card takes over. If the 2nd card fails, the 3rd card takes over.
Thanks
Chris
- 04-04-2004 #2
I know that you can do something called "load balancing" which will use all three at the same time (as a single interface). I've not tried it in Linux yet, but it is possible I know.

I've not, however, heard of anything that does EXACTLY what you're describing.
"Time is an illusion. Lunchtime, doubly so."
~Douglas Adams, The Hitchhiker's Guide to the Galaxy
- 04-05-2004 #3Just Joined!
- Join Date
- Apr 2004
- Posts
- 10
But i did heard about something that called "teaming". Is there a way to have a virtual IP address? Clustering do support virtual IP but requires 2 or more servers. However, i only 1 server to have a virtual IP. How do i do it?
thanks
chris
- 04-15-2004 #4Linux User
- Join Date
- Jan 2003
- Location
- Cardiff, Wales
- Posts
- 478
point
Never heard of teaming.
Network cards don't fail very often. is the app running locally if so why not use the loopback address.
Otherwise you just need to code in somekind of error flag and deal with the error gracefully. There is always a single point of failure. what if all three cards are connected to the same hub and that fails?
Another option is to make your code monitor the eth devices and when one fails switch over to the next available one. Or set your code to use the fqdn and set a script to monitor the eth devices and send an update to the dynamic dns server when one fails.
But this is all very complex for a scenario that shouldn't really happen. You can never really expect software to deal with hardware failure. That's why hardware redundancy is normally catered for using hardware code - bioses and the like.No trees were harmed during the creation of this message. Its made from a blend of elephant tusk and dolphin meat.


Reply With Quote
