Can you update what is VCS version you are using ?
was this setup working before ?
One thing which I observed above is that you are not using Mii to get the health status of interface. In such cases, it is recommended to provide the NetworkHosts attribute so that NIC agent will ping to those IP addresses to determine the health status.
See below section from VCS Bundled Agents guide for 5.0MP3
Monitoring bonded NICs
The NIC agent can monitor the network interfaces (bond0, bond1, etc.) that the
bonding driver exports. Refer to operating system vendor documentation to set
up the bonds and to configure your system to load the bonding driver correctly.
For monitoring a bond interface, the two important settings are:
â– The value of the miimon parameter, which you set while loading the
bonding driver. miimon is a parameter to the bonding module and has a
default setting of 0.
â– The value of the Mii attribute (Mii) of the NIC resource, which you set at
runtime. Mii is an attribute of the NIC resource and has a default setting of
0.
Setting Mii and miimon
For the following cases, the name of the monitored bond interface is B. If you do
not use one of the following cases to set up bonding, the bonding driver can
potentially provide incorrect health status. This incorrect health status can
result in VCS failing to fault the resource appropriately.
Case 1
Accept defaults—miimon is 0 and Mii is 1. Each of B’s slaves must support the
netif_carrier_ok in-kernel call.
Case 2
When you set miimon to anything except 0 (miimon!=0) and Mii to 1, both the
hardware and the drivers of each of B's slaves must support the MII-based
health monitoring.
Case 3
When you set Mii to 0, the NIC agent uses ping, which each card supports. In this
case, the miimon setting is irrelevant.
As you are using case 3, see below the sample configuration:
Sample configurations
Configuration for using Mii
If the NIC does not respond to Mii, the agent uses network statistics to monitor
the device.
NIC groupx_eth0 (
Device = eth0
Mii = 1
PingOptimize = 1
)
Configuration for using network hosts
NIC groupx_eth0 (
Device = eth0
NetworkHosts = { "166.93.2.1", "166.99.1.2" }
)
Here is description of Network hosts:
List of hosts on the network that receive pings to determine the state
of the NIC. Specify the IP address of the host—not the host name.
The specified hosts must be pingable:
â– from all the AppNodes that are specified in the SystemList
attribute for the service group to which the resource belongs
â– through all the devices that are specified in the Device attribute
The command to ping the host (hostip) via a NIC device (nicdev) is:
# ping -I nicdev hostip
If more than one network host is listed, the monitor returns ONLINE if
the ping test is successful with at least one of the hosts.
Type and dimension: string-vector
So I would suggest to populate & come up to the recommeded configuration & see if issue still persists.
Gaurav