Testing for Availability - Expert Oracle RAC Performance Diagnostics and Tuning - page 63

Interrupt:98
eth21:1 Link encap:Ethernet HWaddr 00:23:7D:36:FE:F7
inet addr:169.254.181.158 Bcast:169.254.191.255 Mask:255.255.192.0
UP BROADCAST RUNNING MULTICAST MTU:9000 Metric:1
Interrupt:98
eth22 Link encap:Ethernet HWaddr 00:23:7D:36:5B:90
inet addr:172.30.0.98 Bcast:172.30.0.255 Mask:255.255.255.0
inet6 addr: fe80::223:7dff:fe36:5b90/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:9000 Metric:1
RX packets:665462 errors:0 dropped:0 overruns:0 frame:0
TX packets:168063 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:217845225 (207.7 MiB) TX bytes:113734036 (108.4 MiB)
Interrupt:169 Memory:f8000000-f8012800
eth22:1 Link encap:Ethernet HWaddr 00:23:7D:36:5B:90
inet addr:169.254.3.23 Bcast:169.254.63.255 Mask:255.255.192.0
UP BROADCAST RUNNING MULTICAST MTU:9000 Metric:1
Interrupt:169 Memory:f8000000-f8012800
eth22:2 Link encap:Ethernet HWaddr 00:23:7D:36:5B:90
inet addr:169.254.196.234 Bcast:169.254.255.255 Mask:255.255.192.0
UP BROADCAST RUNNING MULTICAST MTU:9000 Metric:1
Interrupt:169 Memory:f8000000-f8012800
eth23 Link encap:Ethernet HWaddr 00:23:7D:36:5B:94
inet addr:172.30.0.96 Bcast:172.30.0.255 Mask:255.255.255.0
inet6 addr: fe80::223:7dff:fe36:5b94/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:9000 Metric:1
RX packets:88927 errors:0 dropped:0 overruns:0 frame:0
TX packets:81267 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:88117506 (84.0 MiB) TX bytes:29941851 (28.5 MiB)
Interrupt:177 Memory:fa000000-fa012800
eth23:1 Link encap:Ethernet HWaddr 00:23:7D:36:5B:94
inet addr:169.254.110.128 Bcast:169.254.127.255 Mask:255.255.192.0
UP BROADCAST RUNNING MULTICAST MTU:9000 Metric:1
Interrupt:177 Memory:fa000000-fa012800

■ in the preceding output, it should be noticed that there are two haips assigned to eth22. this is because
oracle software assumes that niCs are assigned in pairs and only a max of 4 haips will be assigned for the node.

Note

The Oracle daemon process that manages the HA feature of the HAIP is called Grid IPC daemon (GIPCD).
GIPCD maintains logs in the GRID_HOME/log/<node name>/gipcd directory and watches the availability of the NIC
every 5 seconds, and the GIPCD monitor ( GIPCDMON ) checks on the load average of the NICs every 10 seconds.
If a NIC fails for any reason, the HAIP assigned to that NIC is assigned to one of the surviving NICs (similar to
the database VIP feature that was introduced in Oracle Database 10g Release 2 except that the HAIP does not failover
between servers. It remains on the same node when one or more NICs fail or are disabled when the entire node fails).
When a private NIC fails, the HAIP moves from the failed NIC to one of the surviving NICs on the server; and
when both NICs in the preceding node configuration fails, the HAIP from all the failed private NICs failover to the
surviving NICs. Illustrated in the example output following, the server has four private interconnects configured
using the HAIP feature; three out of the four NICs fail, and the HAIP of all the failed NICs failover to the surviving NIC
( eth21:1, eth21:2, eth21:3 , and eth21:4 ).