NetRAIN declared connectivity alert

[Christof Schoeman]

Hi there

I have recently installed a new system, all on its own on a seperate network. The system has 2 100Mb interfaces, configured in a NetRAIN set (nr0 = ee0 + ee1, ee0 is active).

The network seems fine, but I'm getting thousands of these messages:

NetRAIN: node MYNODE has declared a connectivity alert with network 0.0.0.0 via interface ee1

Same for ee0, from time to time.

and even

...with network 192.168.240.0 via interface nr0

I have managed to reduce the number of messages by setting the nrtimers to 20 and 30, but this is hardly desirable for a NetRAIN set.

Any idea why I am getting these messages, or how I can get rid of them?

Ciao

[Ivan Ferreira]

This could happen if your network adapter and switch ports are configured with auto-negotiation. Disable the auto negotiation on both, using lan_config in tru64 and the respective commands on the switch.

[Al Licause]

What version of Tru64unix and which patch level are you running ?

Which facility it reporting the problem and in which file ?

The connectivity alert is common....it's simply telling you that the inactive device has no assigned address, thus the all 0's address.

If the facility reporting the problem is evm then you can filter these messages. By increasing the timers, you increase the time necessary for NetRAIN to fail to the standby interface in the event of a failure on the active.

Also check netstat -i for any errors on either interface. If they are running clean at 100Mbps fdx then leave autonegotiation alone.

[Christof Schoeman]

Thanks for your quick response.

We're running Tru64 V5.1-3 (aka PK5) with some security ERPs.

The problems are reported by evm. We can filter out some of the event, sure.

Note that we're also getting these events for nr0, which does have an IP address.

autonegotiation is disabled from all sides.

Is niffd not suppose to generate traffic on all the interfaces? If so, it does not seem to happen fast enough, hence the errors. Can one make niffd generate traffic faster?

[Al Licause]

>> Is niffd not suppose to generate traffic on >> all the interfaces?

niffd is the fault finder....not the traffic generator. NetRAIN itself will generate traffic on both interfaces to poll for status.

>> If so, it does not seem to happen fast
>> enough, hence the errors. Can one make
>> niffd enerate traffic faster?

This is a normal informational message though why engineering chose to do this I'm not sure. I would go ahead and filter the messages.

You might want to enable debug on the netRAIN set: ifconfig nr0 debug and see what sort of messages are generated in either daemon.log or kern.log.

[Christof Schoeman]

Thanks again for your comments.

From the niffd(8) man page:

"The Network Interface Failure Finder daemon, niffd, is a traffic generator
for network interfaces that have been classified inactive by the kernel
Traffic Monitor Thread (TMT). The purpose of niffd is to get the interface
packet counters to increment, signifying the interface is still alive and
well."

This is what lead me to believe that it is actually niffd that is perhaps not incrementing the counters fast enough.

I can filter out the messages, but don't want to miss out on a message that indicates a real problem.