Ways to clear ARP entires

Hi All ,

I am having the doubts on the ways to clear ARP queries. Apart from
flush out the arp cache (or) timeout of each entry, is there a way to
clear the ARP entires ?

Consider the Scenario ,

Two machines say ,master and slave plumbing the same floating IP in
the order. Later slave is going down. Because of this none of the
requests are reaching to Master Node on the Floating IP.


Please help me to clear this.


Thanks,
Sasikumar.V

(E-Mail Removed) wrote:
> Hi All ,
>
> I am having the doubts on the ways to clear ARP queries. Apart from
> flush out the arp cache (or) timeout of each entry, is there a way to
> clear the ARP entires ?
>
> Consider the Scenario ,
>
> Two machines say ,master and slave plumbing the same floating IP in
> the order. Later slave is going down. Because of this none of the
> requests are reaching to Master Node on the Floating IP.
>
>
> Please help me to clear this.

Besides arp -d on each client obviously...

Try a broadcast ping to the entire network so everyone will see traffic for
the IP from the new owner and hopefully change their arp cache.

Hello,

Joe Beanfish a écrit :
> (E-Mail Removed) wrote:
>
>> I am having the doubts on the ways to clear ARP queries. Apart from
>> flush out the arp cache (or) timeout of each entry, is there a way to
>> clear the ARP entires ?
>>
>> Two machines say ,master and slave plumbing the same floating IP in
>> the order. Later slave is going down. Because of this none of the
>> requests are reaching to Master Node on the Floating IP.
>
> Try a broadcast ping to the entire network so everyone will see traffic for
> the IP from the new owner and hopefully change their arp cache.

I'm afraid that IP traffic has no effect on the ARP cache. Sending a
"gratuitous ARP" with the new MAC address may help.

Hi guys...

You said:

> Consider the Scenario ,
>
> Two machines say ,master and slave plumbing the same floating IP in
> the order. Later slave is going down. Because of this none of the
> requests are reaching to Master Node on the Floating IP.

I think that ARP cache is not your problem... Consider using a port
monitor daemon...
Thy this: http://sourceforge.net/projects/hapm

It's a good piece of software! When running, both on masters and
slaves, this daemon will check the availability of both and, on a
master fails, the slave will take control of the situation. This is
called HEART BEATING, and is essential daemon on a high availability
project.

May the Force be with you...