NIC agent and Phantom failed

[mokkan]

Hello,

We are running 2 node clusters and all of the sudden NIC  and Phanton agent failed, but servicegroups and resrouces  are working fine. This error is coming from node1 and need to ove the SGs to nod2.  How critical is this error? Do  you think better to restart the Agent?

[mikebounds]

Which resource failed - if it is NIC, then you should check why it has failed. If is the phantom agent then could try to restart the agent or remove phantom agent and recreate it.

The phantom agent should be non-critical so it does't cause the service group to fault.

Mike

[mokkan]

Thank you very much. There is no isse with with NIC, it is the main ethernet interface and there is no error on the card, only it says

Agent Failed

J  NIC  node1

Does it really mean there is a NIC error or can we ignore? The physially network card looks good and interface looks.

Why phantom agent is not critical?

Thanks in advance

[mikebounds]

You need to restart agent using "hagent -start NIC -sys node1"

You should make phantom non-critical because it is not critical for your cluster as it is only a dummy resource.

Mike

[mokkan]

Thank you very much. When NIC agent is faulted, it doesn't mean that there is a problem with NIC right? It may be some thing else as well? IF the NIC agent is faulted, can you switch over the service grup? If we switch over service group from Nod1, do you thing VIPs will be assined to NIC ? Or would it cause problem? There are 5 SGs and all of the them have NIC and IP agents. From my understanding if NIC agent is faulted, IPs can be assigned to new node. Am I right? These are crtical applications and I want to make sure everything is fine.

One more question, by starting agent wouod it cause any issue? From my understanding it won't cause any issue, because we are not touhing the resouce. When the agent reaches the monitor interval, it will keep morning the resource. Am I right?

[mikebounds]

A faulted NIC agent doesn't mean there is a problem with the NIC - agents can fail for other reasons, like system resource issues, but sometimes the NIC agent code comes across something it didn't expect so the agent code fails, so the agent failure MIGHT be related to some unexpected behaviour of the NIC.

The IP resource does not rely on the NIC resource - it doesn't even need a NIC resource to be created, so IPs should failover ok,

You can restart agent without any issues, but you could freeze service group first, just incase when NIC agents restarts, it things NIC is down for some reason.

Mike

[mokkan]

Hi Mike,

Thank you very much for your well explanation.

I have a question regaring your following statement:

"The IP resource does not rely on the NIC resource - it doesn't even need a NIC resource to be created, so IPs should failover ok"

You said IP resource doesn't realy on the NIC resource, if NIC resouce failed due to NIC issue, how do we setup the IP Address? How IP resource is going to online? We already seutp the depencies between NIC and IP.

One more thing. IF we stop NIC Agent, would NIC status would show as offline or unown?

[mikebounds · Accepted solution]

NIC agent monitors the network, so if for instance you unplugged the network cable, the NIC resource would fail as the host would not be able to ping out from the NIC, but in this scenario you can still add a virtual IP to the NIC.  Bu there is no point putting a VIP on a NIC that doesn't work, hence why you should have NIC and IP resources, but the point is if the NIC agent is incorrectly saying there is an issue with the NIC, then this will not effect the IP resource working.

I believe if the NIC agent fails, then the NIC resources show their last known state.

Mike

[mokkan]

Thank You very  much Mike.