Cisco Support Community
cancel
Showing results for 
Search instead for 
Did you mean: 
Announcements

Welcome to Cisco Support Community. We would love to have your feedback.

For an introduction to the new site, click here. And see here for current known issues.

New Member

Virtual IP PING behaviour is conflicting as the VIP is not pingable but application is able to connect to the database on the rebooted node.

Problem :

----------

we have a Customer which has  2 node RAC 10.2.0.4 running on RHEL 5 on x86_64 DB oracle servers. The application was not able to failover automatically to the

surviving node .

Observation :

-----------------

1. Database instance was shutdown on 1 node . Application successfully fails over.

2. Clusterware stack was shut down on 1 node . Application successfully fails over.

3. One node shut down . Application unable to failover.

Analysis :

-----------

After server shut down , the failed over VIP is not pingable .

As per Oracle , this was the issue with one of the bonding modes (balance-alb ) used for NIC card bonding . It was confirmed that the same mode was used.

Oracle highlights this is not a database issue and other bonding modes should be tested.

Workaround:

----------------

The bonding mode was changed and after several testing with different bonding modes ( 1, 3, 5 ) , we finally settled with bonding mode 1 (active backup ) . This is used specifically once multiple switch topology is used .

Observation:

---------------

1. The VIP fails over to surviving node if the parent node is shutdown . Accepted behavior.

2. The failed over VIP is now pingable. Earlier it was not pinging.

3. Application not able to failover automatically . Same as earlier.

4. Once node is brought back , the VIP fails back . Accepted behavior.

5. The failed back VIP is still not pingable.

6. The Application is able to connect to the recently brought back node.

Analysis:

-----------

1. As per point 5 and 6 , the PING behaviour is conflicting as the VIP is not pingable but application is able to connect to the database on the rebooted node.

Can any one suggest what need to check as part of this ping issue..this server are behind FWSM one of the server segment..and we have applied access list to inside segment to  restricted the access and outiside allowed all.

This servers are connected to 4948 switch with normal access port configuration.

This server physical and VIP are in same subnet..so routing procespctive in FWSM nothing has to be done.

Regards

-Saji

839
Views
0
Helpful
0
Replies
CreatePlease login to create content