lan card

[Ragni Singh]

Hi All,

this morning I was unable to ping one of my interfaces. I did a ifconfig on the lan card and it showed as being up. I then did a lanscan and it showed as being down. I went into lanadmin and did a reset of the interface and at that point the interface came up. Looking at the syslog didn't tell me anything about what time is happened or why?. Would I have seen this hardware failure in any logs and if yes, which log.

Can openview notify me of any hardware failures as such. Any help will be greatly appreciated and points will be assigned.

[James R. Ferguson]

Hi:

My guess would be that your card is set to allow auto-negotiation and that connectivity may have dropped for a very short interval. My experience has been that usually '/var/adm/syslog/syslog.log' records a lan failure, though.

Regards!

...JRF...

[Ron Kinner]

I think IFCONFIG just tells you that the NIC has been put into service from a software point of view. Look at
lanadmin
lan
display
and it usally shows two ups. One of them is the equivalent of the IFCONFIG up the other tells you if it's really working.

Assuming OpenView's node manager is pinging your IP address you should see it turn red when it doesn't answer. Might take up to 5 minutes to notice it tho unless you have a second NIC and it can send a trap out the second NIC to tell you the bad news.

Ron

[Christopher McCray_1]

Hello,

Also /var/adm/messages can have this information. In my /etc/syslog.conf file I have:

*.emerg *

Which, among the syslog and console, will write out to /var/adm/messages.

The dmesg command should also report this information.

Hope this helps

Chris

[Kellogg Unix Team]

Also, check /var/adm/nettl.LOG00 or nettl.LOG01

netfmt /var/adm/nettl.LOG00

...Manjeet

[Keith Clark]

Hi Sanjay,

A year or two ago we had intermittent problems with our 1000-BaseSX cards after applying a patch. The card would lose either transmit or receive (I can't remember which), but from a ServiceGuard perspective still had link. HP was quick to identify the problem and we rolled back the patch, but we were uncomfortable with the fact that this could happen and our users would know before we did.

We setup a BMC Patrol KM that uses HP-UX's nettl tracing to monitor the cards. Openview should be able to do the same thing.

The BMC knowledge module runs the netfmt command using a filter file and in "Follow" mode. It basically greps out the headers and any output detected after that cause an alarm and a page.

We use /usr/sbin/netfmt -c -f -F /var/adm/nettl.LOG00

The filter file (example attached) is setup to monitor only the GELAN subsystem. We had to play around with the filter file a bit to get only what events we wanted, but it works quite well. We have even detected faulty GBICs in switches before ServiceGuard has had time to react. It was well worth the effort to set this up.

Good luck,

Keith