MUltiple IP creation with IP redundancy

[sharpsanjay]

Hi,
I have to implement hardware redundancy.I have 2 linux machine for this and each have similar configuration/hardware. Lets say two machine Linux1 and Linux2. Currenty Linux1 have one NIC and it has 4 Ip address bind on it. Now I have to down the Linux1 and up Linux2 with same Ip addresses.
Could it possible to configure all the Ip (same)address on both Linux1 and Linux2 and Linux1 have eth status UP while Linux2 have eth:y status down.
I am trying it but on Linux2 machine I cant able to create multiple Ip interfaces with down status. Could you please suggest me regarding this.
Thanks a lot.

[Lazydog]

If I understand your question correctly the answer is no.
The sub-interfaces are in the same state as the main interface.
So if eth0 is up then eth0:x are also up.  The only way I know of 
to have eth0 up and all or a few eth0:x down is to remove the 
sub-interfaces you want to be in a down state from eth0.

Sorry for the code tags, haven't figured out how to turn off the smileys.

[sharpsanjay]

my requirement is kinda server redundancy. Linux box one have 2 NICs and 6 IP Address and Linux BOX2 have 2 NICs and same 6 IP to be use in case of Linux Box1 crashes/fails. Box1 and Box2 have two different IPs for health monitoring of each other rest of the IP should be act as backup to each other.

What cud be possible way to swap/switch the linux boxes with min time. should i stored all IP of Linux box1 onto Linux box2 and in case of box1 failure, trigger the Ip creation on box2? Is there any better mechanism so that we configure all ip on both unit but on box1 status is up and box2 have status down for those IPs and in case of switch over simple set the status up/down without all stuffs like VLAn/route..

[Lazydog]

This is something that you will have to decide what works best for you. Another thing to think about are is the fail over going to be automatic or manual? There are other ways of doing this without 2 machines. you need to look at what might stop a system from responding.

Hard drive failure (easy to work around)
NIC card failure (easy to work around)
Network failure (depends on where the failure is if it can be worked around)
Mother board failure (dead machine)

For hard drive failure simply run a RAID configuration. If you lose a disc another one picks up.

NIC card failure you just need to have a second nic on the same subnet and configure it, which your machine already has a second nic.

Network failures on the internal LAN are rare. simple fix is to have a spare part. Now if the network fails on the WAN you have an issue as you normally don't control this part and must wait on the vendor to fix it. Here a second server might come in handy but the ip address you can use are not going to match the first machine as it will be on a different network.

Mother board failure is one that there is nothing you can do but have a second box ready. Mother board failure are few and far between.

I would use one box with the above fixes instead of 2 systems. It is simple and it is not as expensive as 2 systems to maintain.

[sharpsanjay]

its prototyping to simulate some real time operation of tele communication.
The need of such configuration is to make the redundant unit.so that in case of up gradation of BOX1, box2 can take charge either manually or automatically(by health monitoring) and vice varsa.
y i used multiple IP on same physical port is due to differentiate the nature of data.
I am using red-hat for simulation but finally I have to implement it on some MIPS processor.
I configured box 1 with multiple virtual IPs, VLAN, and route addition. I required smiliar configuration on box2 but box2 shud not be know by other then box one(for this box1 and box2 have seperated IP for health monitoring). At the time of swtichover I want it fast . cud it be possible in linux to attached IPs , VLAn, routed but in down status.. and as soon box2 gets command to lead it only up the virtual inferfaces without geeting list of Ips etc.. and making configuration again.
I am beginnerin this and will be thankful for ur expert comment ...
-San