topic Re: Oracle heartbeat problem in Operating System - Linux

Oracle heartbeat problem

Timothy Nibbe — Tue, 07 Dec 2010 14:20:05 GMT

We have four servers that are in a four node Oracle ASM cluster. We are using two NICs; the first NIC is the server connection with an MTU of 1500, and the other is for the cluster heartbeat, which has an MTU of 9000 (it is not bonded). The first NICs are connected to our LAN, and the second NICs are connected to a private VLAN.

The first NICs are operating normally, the heartbeat NICs occasionally drop the connection; the server with the failed NIC is unable to ping to the other three servers and the other three servers are unable to ping to the server with the failed NIC. THe failed NIC stays failed, however the NIC can be put back into service by restarting the network.

Re: Oracle heartbeat problem

Alzhy — Tue, 07 Dec 2010 15:57:42 GMT

What kind servers and what kind NICs do yo have? Is there a reason why your MTU needs to be at 9000 for the heartbeat ?

Re: Oracle heartbeat problem

Tim Nelson — Wed, 08 Dec 2010 18:04:30 GMT

I would ask the same question.

why set MTU on HB lan to 9000 ?

set back to default of 1500 and see if your issue goes away ..

Re: Oracle heartbeat problem

Tim Nelson — Wed, 08 Dec 2010 18:07:21 GMT

here is a great overview that may help.
http://en.wikipedia.org/wiki/Maximum_transmission_unit