IPMP test address selection
hello everyone. i have dedicated network with two servers with solaris9 configured. sunray servers installed on both. both servers connected by two NIC's with IPMP configured. there are no anymore devices in the network. i have following trouble: if any of servers down (reboot, halt etc) anothers server stop answer on network request with message in log: 'in.mpathd[236]: [ID 168056 daemon.error] All Interfaces in group test have failed'. all becoming normal, when rebooted server online. i got log message:
Jul 4 11:59:10 sf210-contourK in.mpathd[236]: [ID 299542 daemon.error] NIC repair detected on bge3 of group test
Jul 4 11:59:10 sf210-contourK in.mpathd[236]: [ID 237757 daemon.error] At least 1 interface (bge3) of group test has repaired
Jul 4 11:59:10 sf210-contourK in.mpathd[236]: [ID 620804 daemon.error] Successfully failed back to NIC bge2
Jul 4 11:59:10 sf210-contourK in.mpathd[236]: [ID 299542 daemon.error] NIC repair detected on bge2 of group test
snoop show, that avery server selectig addresses of opposite server by test addresses:
(10.99.0.1-2-3 - one server, 10.99.10.1-2-3 second server)
10.99.0.2 -> 10.99.10.3ICMP Echo request (ID: 60421 Sequence number: 5594) 10.99.10.3 -> 10.99.0.2ICMP Echo reply (ID: 60421 Sequence number: 5594)
10.99.10.1 -> 10.99.0.3ICMP Echo request (ID: 65027 Sequence number: 5595)10.99.0.3 -> 10.99.10.1ICMP Echo reply (ID: 65027 Sequence number: 5595)
10.99.0.1 -> 10.99.10.3ICMP Echo request (ID: 60420 Sequence number: 5595) 10.99.10.3 -> 10.99.0.1ICMP Echo reply (ID: 60420 Sequence number: 5595)
looks like this is reason by total network fail. is any solution for solving rouble like this?
