cancel
Showing results for 
Search instead for 
Did you mean: 

VRTSvcs NIC:Nic:monitor:Invalid Device Name

mrossini
Level 3

Hello.

I've a Veritas Cluster Server 5.0MP3RP3 on Solaris 10 with a MultiNICA & NIC resource configured.

The host "indirizzalo" is configured with bge0 and nxge1.

 

group net-sg (
SystemList = { consegnalo = 0, trasportalo = 1, accettalo = 2,
indirizzalo = 3 }
Parallel = 1
AutoStartList = { consegnalo, trasportalo, accettalo }
)

MultiNICA MultiNicA (
Critical = 0
Device @consegnalo = { bge0 = "192.168.45.81", nxge1 = "192.168.45.81" }
Device @trasportalo = { bge0 = "192.168.45.82", nxge1 = "192.168.45.82" }
Device @accettalo = { bge0 = "192.168.45.43",
nxge15 = "192.168.45.43" }
Device @indirizzalo = { bge0 = "192.168.45.135",
nxge1 = "192.168.45.135" }
)

NIC Nic (
Critical = 0
Device @consegnalo = bge0
Device @trasportalo = bge0
Device @accettalo = bge0
Device @indirizzalo = bge0
NetworkType = ether
NetworkHosts = { "192.168.45.50" }
)

Phantom Phantom (
Critical = 0
)

 

Recently I had a network fail on the bge0 interface.

The cluster switched correctly to the nxge1 interface but now the 'NIC' resource fails the probe since it probe the bge0 interface instead the nxge1

 

# ifconfig nxge1

nxge1: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 5
inet 192.168.45.135 netmask ffffff00 broadcast 192.168.45.255
ether 0:21:28:8:dd:8f

# ifconfig bge0

ifconfig bge0
ifconfig: status: SIOCGLIFFLAGS: bge0: no such interface

 

when an interface is not active it is unplumbed (seeing the other nodes, that have bge0 plumbed and nxge1 unplumbed).

The cluster log:

2016/11/07 10:18:29 VCS WARNING V-16-10001-7502 (indirizzalo) NIC:Nic:monitor:Invalid Device Name
2016/11/07 10:18:29 VCS WARNING V-16-10001-7509 (indirizzalo) NIC:Nic:monitor:Device name (bge0) could be a typo or Device is not configured.Resource State is OFFLINE!
2016/11/07 10:18:30 VCS INFO V-16-2-13001 (indirizzalo) Resource(Nic): Output of the completed operation (monitor)
ifconfig: status: SIOCGLIFFLAGS: bge0: no such interface

 

Currently the resource is in freeze state to avoid unknown faults.

why the cluster does not probe the active interface? There is some misconfiguration?

Can I switch active interface from nxge1 to bge0 since the network fault was solved?

 

Thankyou

Matteo

 

 

 

4 REPLIES 4

RiaanBadenhorst
Moderator
Moderator
Partner    VIP    Accredited Certified

Hello

Your SG has both NIC and MultiNICA configured. The NIC is configured for bge0 so what the cluster is saying is that if cannot find this interface in the OS. You should choose one or the other, not both.

bge0: no such interface

I would make sure that the failure is properly resolved. You can try and plumb bge0 and maybe put an IP on it to make sure.

mmm.

we have this configuration from years...

 

I tried to plumb the interface, but without put an ip address.

There are some risk by setting an ip address on an interface configured in cluster? If will occour a fail, it will switch properly?

How I should reconfigure the cluster? needs to stop services?

RiaanBadenhorst
Moderator
Moderator
Partner    VIP    Accredited Certified

Hi,

 

Plumbing the interface should make it visible to the OS and VCS will see it. You don't need to have an IP on a NIC for the NIC monitor to report it as online (unless you maybe configure a network host for it to ping)

I saw the code of the monitor script.
It checks for the interface plumbed and UP.
As you saw, should not be both multinica and NIC resources in a cluster configuration.
However both resources are configured with Critical = 0, so the only problem is the message in the log.

The most clear and definitive solution should be to remove the NIC resource and just leave the MultiNicA resource.
But the most simple and faster and secure solution is to reconfigure the resource to check the nxge1 interface.
It is not a definitive solution and fails after a reboot, but surely it is better than manually plumb and up the bge0 interface.

Thank you.