Re: Oracle RAC and VIPs

From: Nadeem M. Khan <nadeem.m.khan_at_gmail.com>
Date: Thu, 10 Jul 2008 22:28:23 +0400
Message-ID: <2ede0e110807101128s1bc92c84uc1aca82448711503@mail.gmail.com>

On Thu, Jul 10, 2008 at 9:03 PM, Baumgartel, Paul <paul.baumgartel_at_credit-suisse.com> wrote:
> If the VIP comes up on another node, doesn't the attempt to connect to that
> (virtual) IP translate to a connect to the other node? If not, how exactly
> is the failure detected right away?

When the VIP fails over to another node, it is not supposed to belong to that node. The MAC address and listener configuration of that node do not match the VIP. This is detected by the client immediately and a connection reset (negative ack) occurs. The client then tries to connect to the next available node.

If there was no VIP, the client would have taken around 10 mins to detect a connection timeout as there would have been no reset signal. This is to say that the purpose of the VIPs is not just failover, but *fast* failover.

Regards,
NMK.

--
http://www.freelists.org/webpage/oracle-l

Received on Thu Jul 10 2008 - 13:28:23 CDT

This message: [ Message body ]
Next message: J. Dex: "Solaris SPARC or x86?"
Previous message: Lion, Oren-P64304: "RE: Solaris SPARC or x86?"
In reply to: Baumgartel, Paul: "RE: Oracle RAC and VIPs"
Next in thread: Alessandro Vercelli: "Re: Oracle RAC and VIPs"

Contemporary messages sorted: [ by date ] [ by thread ] [ by subject ] [ by author ]

Original text of this message