asked on

heartbit with eth0:0

hi ..

I need an idea ...i want exactly what i need as i described bellow ..

I need to setup nfs share in to server (Server A and Server B ) so that Server C can access it via ip 192.0.0.4 ( which will move round beween Server A and SErver B)

Both server has different Ip example (Server A : eth0 192.0.0.2 , Server B:eth0 192.0.0.3)

I will setup Same nfs share on both server.

initially i will create eth0:0 in SErver A with ip 192.0.0.4

and heatbit program will be on both server .

When Server A will go down ..

I want to map 192.0.0.4 from SErver A , eth0:0 to SErver B ,Eth0:0

can it be done ??

I belived it can be done and i am working on

it any one wants to advise me something.. please advise

Thanks for your help.

Server C

ASKER CERTIFIED SOLUTION

arnold

membership

This solution is only available to members.

To access this solution, you must be a member of Experts Exchange.

Start Free Trial

fosiul01

ASKER

Im Using debian,.,.

arnold

I've not setup clustering with debian.

You could setup a script on each node. That will first check whether it is the primary (handling the IP in question)
http://www.linuxquestions.org/questions/debian-26/best-way-to-cluster-with-debian-553908/

The 192.0.0.4 will by the IP resource that will follow to the active node.

Centos/RedHad have the system-config-cluster that simplifies the setup.

Are you looking to manage it manually/through scripts?
The issue is that you have to deal with RACE conditions to avoid attempting to bring the IP on both nodes.

arnold

http://www.debianadmin.com/how-to-set-up-a-high-performance-cluster-hpc-using-debian-lenny-and-kerrighed.html

If you do use script, you would use ifconfig eth0:0 <ip> netmask <netmask> up
to bring up the IP on the correct interface.
To manage the arp expiration delay, you may want to specify/use a common mac address in the script.

fosiul01

ASKER

About the race condition... I dont know yet because im using oracle vm and its shared storage just to
Make a cluster as workway round and im using
Nfs for this shared storage . Only problem is if
Nfs server goes down VM will shutdown the server.
So i need HA to quickly move the ip.

Does heartbeat work with normal debism?

arnold

I do not know. Any specific reason you are using debian versus centos/redhat?

I would think they have to have a heartbeat test which is how the nodes communicate and determine which one is active and which is the standby and when communications are lost to the active, the standby asserts bringing up the resources that are part of the cluster.

You mean oracle Linux VM.
The VM will likely go down because of the transition is not instantaneous for the IP resource. The access to the file within NFS will not be instantaneous either.

IP is likely a 5 second transition. The NFS related lock acquire access to the file is an additional time overhead.

fosiul01

ASKER

yes Oracle Vm 3

this version require to have atleast one Cluster (shared storage)

and we dont have any central storage..

so i have decided to go with nfs..

for test environment , i used openfiler which works fine..

but due to resources.. in live environment, i cant use openfiler..

all our servers are debian.. so my manager said, to use any existing debain server which is in live environemnt and make nfs share..

our server is very reliable but you never know when one goes down

so i cant take any risk of having only one nfs share..

so I need to put that share in to another server incase if main one down... ip will go over to next one and ORACLE VM will not cause any problem...

i asked question in Oracle vm to see how often the OVM manager check for live storage..

i used before heartbeat + drbd..

but i am thinking to use just hearbeat so that it move around the iP ...

but dont know if it can be done with pure debain or not ..

fosiul01

ASKER

i think it will work

http://www.howtoforge.com/high-availability-load-balancer-haproxy-heartbeat-debian-etch-p2

but need to test it ...