This discussion is archived
4 Replies Latest reply: Jan 29, 2013 1:14 AM by Sebastian Solbach (DBA Community) RSS

VIP network of RAC are on Node 2 Only

901518 Newbie
Currently Being Moderated
Hello,

This is related to this thread: https://forums.oracle.com/forums/thread.jspa?messageID=10817046#10817046, which is (as of writing) is still unresolved.

Is it normal to have the VIP eth are on the one node only?
Below is the result of ifconfig on Node 2:

[root@bsspbbi2 ~]# ifconfig
eth0 Link encap:Ethernet HWaddr 00:50:56:A9:D6:3D
inet addr:192.168.216.36 Bcast:192.168.216.255 Mask:255.255.255.0
inet6 addr: fe80::250:56ff:fea9:d63d/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:7535526 errors:0 dropped:0 overruns:0 frame:0
TX packets:25017957 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:43105775163 (40.1 GiB) TX bytes:31219340971 (29.0 GiB)
Interrupt:10 Base address:0x2800

eth0:1 Link encap:Ethernet HWaddr 00:50:56:A9:D6:3D
inet addr:192.168.216.45 Bcast:192.168.216.255 Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
Interrupt:10 Base address:0x2800

eth0:2 Link encap:Ethernet HWaddr 00:50:56:A9:D6:3D
inet addr:192.168.216.46 Bcast:192.168.216.255 Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
Interrupt:10 Base address:0x2800

eth1 Link encap:Ethernet HWaddr 00:50:56:A9:70:51
inet addr:192.168.215.36 Bcast:192.168.215.255 Mask:255.255.255.0
inet6 addr: fe80::250:56ff:fea9:7051/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:1890956 errors:0 dropped:0 overruns:0 frame:0
TX packets:1025292 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:1812303471 (1.6 GiB) TX bytes:514278872 (490.4 MiB)
Interrupt:5 Base address:0x2840

lo Link encap:Local Loopback
inet addr:127.0.0.1 Mask:255.0.0.0
inet6 addr: ::1/128 Scope:Host
UP LOOPBACK RUNNING MTU:16436 Metric:1
RX packets:60508 errors:0 dropped:0 overruns:0 frame:0
TX packets:60508 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:6583165 (6.2 MiB) TX bytes:6583165 (6.2 MiB)

------

192.168.216.45 is the VIP of Node 1, and 192.168.216.46 is the VIP of Node 2.

When the system starts up, both nodes have the eth0: 1 network. But after sometime, it seems to me that the "virtual network cards" (I do not know the right term), are being transferred between two nodes. And when that happens, either the listener of node 1, or the listener of node 2, or both are on "OFFLINE" status..

Our data import and loading activity are greatly affected by this issue, and we are now really far from schedule.
Please help.

Thanks.
  • 1. Re: VIP network of RAC are on Node 2 Only
    Sebastian Solbach (DBA Community) Guru
    Currently Being Moderated
    Hi,

    unfortunately you did not mention which Oracle version this is.
    It seems your public network check is failing, hence the VIP gets relocated to the other node (where it still seems to be o.k.).
    Depending on the used version, the VIP will not fallback by itself.

    Why does the public network check fail?
    Again there are different mechanism used per Oracle DB version.

    Normally it indicates a problem with a network card, but since this is happening on both nodes, I more likely suggest, that you have a problem with the gateway (that it is temporarily not reachable), resulting in the failure of the VIP.

    Regards
    Sebastian
  • 2. Re: VIP network of RAC are on Node 2 Only
    damorgan Oracle ACE Director
    Currently Being Moderated
    Can you identify by manufacturer and model the following?

    1. NIC cards in use
    2. Switch to which they connect
    and
    3. IP address of cache fusion interconnects
  • 3. Re: VIP network of RAC are on Node 2 Only
    901518 Newbie
    Currently Being Moderated
    Sebastian Solbach (DBA Community) wrote:
    Hi,

    unfortunately you did not mention which Oracle version this is.
    It seems your public network check is failing, hence the VIP gets relocated to the other node (where it still seems to be o.k.).
    Depending on the used version, the VIP will not fallback by itself.

    Why does the public network check fail?
    Again there are different mechanism used per Oracle DB version.

    Normally it indicates a problem with a network card, but since this is happening on both nodes, I more likely suggest, that you have a problem with the gateway (that it is temporarily not reachable), resulting in the failure of the VIP.

    Regards
    Sebastian
    Hi Sebastian

    Thanks. This is actually related to my post here: Listener Suddenly Stops (Oracle DB 10g R2 RAC on VMWare)
    Please find time to read that, maybe you can see the details that you need and maybe from there you can suggest.

    Thank you very much for the help. :)
  • 4. Re: VIP network of RAC are on Node 2 Only
    Sebastian Solbach (DBA Community) Guru
    Currently Being Moderated
    Hi,

    in 10.2.0.4. the public network check is done by trying to ping the gateway. If this is not reachable (also if it is only for a short time), then the VIP will failover to the other node.
    In 10.2.0.4. the default is to not failback the VIP, since this will cause another "outage" on the VIP. However you could reenable that in the VIP script.

    See the following notes:

    The listener stops or goes offline randomly due to VIP issue (Doc ID 985170.1)
    10g/11gR1: VIP does not relocate back to the home node even after the public network problem is resolved (Doc ID 805969.1)
    VIPs Often Go Offline Unexpectedly and Relocate to Another Node (Doc ID 1297867.1)

    Regards
    Sebastian

Legend

  • Correct Answers - 10 points
  • Helpful Answers - 5 points