Network interface shutdown due to network disconnect

[George D. Hodge]

Hello all,

The problem I am experiencing probably has been dealt with here before but I have been unsuccessful in my search attempts.

I administer four C240 workstations running HP-UX 10.20. Every Sunday at 11:15 am three of the four machines log a network disaster message saying thatHPCORE 10/100BASE-T driver detected bad cable connection between the adapter in slot 0 and the hub or switch. Sometime after that the network interface is shutdown and the following morning people can't access their accounts via NFS.

Is there anyway to turn off the shutdown of the network interface? Resetting of the network interface using lanadmin from cron is not a possibility if I read the man page for lanadmin correctly. Any help would be appreciated.

Thanks,

George Hodge

[Pete Randall]

Hi George,

First off, I'd be extremely curious why this is occuring at the same time on the same day every week. I would think you might be able to correct the cause rather than patch the aftereffect. Secondly, I don't think your interface should be shutting down - how are your patch levels? Thirdly, and only as a last resort, you could always set up a weekly re-boot of the workstations at 11:30 via cron.

Good luck,
Pete

[harry d brown jr]

George,

Is it possible that your router is being rebooted during this time?

The only reason for this error is when someone unplugs the cables, or the router fails/reboots. This is NOT system related.

live free or die
harry

[Tim D Fulford]

That is weird.. I would have said that the lan card is duff or the cable has been removed. You get a similar message when removing a cable.

I assume you get the intyerface backup up by doing "ifconfig lan up" ?

Tim

[Stefan Farrelly]

Ive seen this problem before. I dont think there is anything you can do on your HP workstations/servers to prevent the Lan driver from 'shutting down' due to network disconnection. We had to find the source of the network problem (usually someone rebooting a switch or hub etc.) and stop that happening, or reboot the HP box to reset the driver.
In my experience this is only the 100Mb lan driver which seems succeptible to this, the old 10MB network cards never seem to suffer from 'driver shutdown'.

If you cant find out who is messing up the network at such a routine time and get them to stop it then you could automatically reboot the HP's after that time to reset/reload the lan driver.

[Pete Randall]

This reminds me of the old story where they discover that the cause of the regular shutdown is the janitor unplugging the server/workstation/router/whatever in order to plug in their vacuum cleaner.

;^)

Pete

[Stefan Farrelly]

I just recalled there was a possible way to make the 100MB lan drive more stable (stop it shutting down due to too many network disconnects), that is make its port on its hub/switch connection 100MB HD/FD. ie. hard set it, dont make it autosensing. If your port is set to autosensing then try making it 100HD/FD.

[George D. Hodge]

Thanks for the replies everybody. I have our network administrator looking at the hub and switch issues. I just need something for the mean time. I run lanadmin manually Monday morning to reset the interfaces. Rebooting every Sunday is not an option as users sometimes have programs running over the weekend.

When I talked to HP about this a few years ago when I had an occurrance of this problem, they said nothing could be done about it and that it's just the way of the kernel. I had some 725/50s running HP-UX 9.03 that never exhibited this behavior so I'm guessing they put the behavior in HP-UX 10.X

I would just be nice if there was something I could do to turn off this behavior. I found a script put forth by Patrick Wallek to run lanadmin. I may have to try that from the cron until the network folks can figure it out.

Thanks,

George

[James R. Ferguson]

Hi George:

Interesting. I've seen the same thing as Stefan. I have have a K370 server running a well-patched 10.20 version and a btlan1 (HP-PB 100 Base) card that occassionally suffers this when there is a "burp" in a switch or router on the network.

I'll usually see the infamous 'LAN disaster' message on the console. Worse, however, is that the LAN interface becomes catatonic. I can't ping out of or into the server. A reboot is necessary to cure the ill.

Regards!

...JRF...

[George D. Hodge]

James,

You might want to try the lanadmin command to avoid rebooting. There is a provision to reset the network interface. It works like a charm for me.

George

[James R. Ferguson]

Hi (again) George:

Aahhh! I had not thought about 'lanadmin's reset option. I'll certainly try that the next time. The occurances of what I've described are occassional, but any unplanned reboot is one too many. Many thanks for the advice!

Regards!

...JRF...