racnode1 APA group Public VIP SCAN IP Cluster Interconnect S.

No 1 Network Interface lan900 lan900:801 lan900:803 lan901 Test Scenario

Failover within public lan Auto Port Aggregation(APA) Bonding

2

Failover within public lan Auto Port Aggregation(APA) Bonding

3 Failover within cluster Interconnect lan Auto Port Aggregation(APA) Bonding

4 Failover within cluster Interconnect lan Auto Port Aggregation(APA) Bonding

5

Public Network Failure

2 and higher) .2.6 Public Network Failure 7 Interconnect Network Failure (11.0.

2 and higher) .8 Interconnect Network Failure (11.0.2.

1.racnode1 IP Address 172.20.20.20.20.1 172.20.1.1 .20.1.36 Action Removed cable 1 of PUBLIC lan900 Connected back cable 1 of PUBLIC lan 900 Removed cable 2 of PUBLIC lan900 Connected back cable 2 of PUBLIC lan 900 Removed cable 1 of PUBLIC lan900 Connected back cable 1 of PUBLIC lan 900 Removed cable 2 of PUBLIC lan900 Connected back cable 2 of PUBLIC lan 900 Removed cable 1 of CLUSTER INTERCONNECT lan901 Connected back cable 1 of CLUSTER INTERCONNECT lan 900 Removed cable 2 of CLUSTER INTERCONNECT lan901 Connected back cable 2 of CLUSTER INTERCONNECT lan 900 Removed cable 1 of CLUSTER INTERCONNECT lan901 Connected back cable 1 of CLUSTER INTERCONNECT lan 900 Removed cable 2 of CLUSTER INTERCONNECT lan901 Connected back cable 2 of CLUSTER INTERCONNECT lan 900 172.1 Unplug all network cables for public network of node 1 (172.1.1.1.1.20.20.5 192.1.1.20.3 172.1.1 racnode2 172.1) 172.1.2 172.2 APA group Public VIP SCAN IP Cluster Interconnect Host 172.

1 .2) 172.Unplug all network cables for public network of node 2 (172.20.20.1.1.2 Unplug all network cables for the interconnect network of node 1 172.20.1.

Unplug all network cables for the interconnect network of node 2 172.1.2 .20.

40 Observed Results Status SUCCESS Failover of IP within in APA.1. * VIP was in INTERMEDIATE status although it failed over to node 2( This is expected behavior starting from 11g) * Database instance was running and unregistered from remote listener.*. o SCAN VIPs and SCAN LISTENERs running on the node will fail over to a surviving node. SUCCESS .20. • Database services will fail over to one of the other available nodes.network and listener resources will go offline for the node.racnode2 Network Interface lan900 lan900:801 NA lan901 Expected Results IP Address 172. clients should fail over to an available * ora. o The VIP for the node will fail over to a surviving node. • The database instance will remain up but will be unregistered with the remote listeners. No Failover of IP within in APA. * TAF connections with uncommitted transactions failed over to node 2 as expected.1. • If TAF is configured.1.*.network and listener resources went offline as expected * SCAN VIP and SCAN LISTENER failed over to node 2 as expected.4 NA 192.2 172. No Failover of IP within in APA. No impact to Oracle impact to Oracle SUCCESS Check with “crsctl stat res –t” o The ora.20. No impact to Oracle impact to Oracle SUCCESS SUCCESS Failover of IP within in APA.1.

2 and above: • CSSD will detect split-brain situation and perform one of the following: o In a two-node cluster the node with the lowest node number will survive. a graceful shutdown of Oracle Clusterware will be attempted. OHASD will attempt to restart the stack. If reboot less eviction fails for some reason. o All I/O capable client processes will be terminated and all resources will be cleaned up. * VIP was in INTERMEDIATE status although it failed over to node 2( This is expected behavior starting from 11g) * Database instance was running and unregistered from remote listener. * Since this is 2 node cluster.2. o In a multiple node cluster the largest sub-cluster will survive. as starting from 11g. clients should fail over to an available * ora.network and listener resources went offline as expected * SCAN VIP and SCAN LISTENER failed over to node 1 as expected. eviction type is reboot less. Node was not rebooted. the 2nd node got evicted from the cluster as expected. * TAF connections with uncommitted transactions failed over to node 1 as expected. In this case the stack will be restarted once the network connectivity of the private interconnect network has been restored. database instance and ASM were abnormally shutdown. • If TAF is configured. * VIP was in INTERMEDIATE status although it failed over to node 2( This is expected behavior starting from 11g) * After reconnecting the private interconnect cables. crs stack and resources started. * Cluster services on node 2 went down. SUCCESS . SUCCESS For 11. then node will be rebooted. If process termination and/or resource cleanup does not complete successfully the node will be rebooted. o SCAN VIPs and SCAN LISTENERs running on the node will fail over to a surviving node.*. o Assuming that the above has completed successfully. • Database services will fail over to one of the other available nodes.network and listener resources will go offline for the node. • On the node(s) that is being evicted. o The VIP for the node will fail over to a surviving node. • The database instance will remain up but will be unregistered with the remote listeners.*.Check with “crsctl stat res –t” o The ora.0.

a graceful shutdown of Oracle Clusterware will be attempted. o All I/O capable client processes will be terminated and all resources will be cleaned up.2 and above: • CSSD will detect split-brain situation and perform one of the following: o In a two-node cluster the node with the lowest node number will survive. OHASD will attempt to restart the stack. o Assuming that the above has completed successfully. * Since this is 2 node cluster.0. database instance and ASM were abnormally shutdown * After reconnecting the private interconnect cables.For 11. * Cluster services on node 2 went down. • On the node(s) that is being evicted. the 2nd node got evicted from the cluster as expected. In this case the stack will be restarted once the network connectivity of the private interconnect network has been restored. If process termination and/or resource cleanup does not complete successfully the node will be rebooted. o In a multiple node cluster the largest sub-cluster will survive. some of the cluster daemons started but ASM and Database did not start automatically( Not as expected) * VIP Failed over but was in INTERMEDIATE state(not as expected) * DB sessions connected using TAF over VIP and not having uncommitted transactions failed over to node 1.2. SUCCESS .

Sign up to vote on this title
UsefulNot useful