HP 9000 and T-1 problems

[penguin1]

We have an HP 9000 UNIX server and a T-1 that connects it between the main office and a factory several miles away. Once in a while, the T-1 goes down and anyone who was on the server becomes "blind" to the server unless we power down the server completely and then bring it back up. Is there a way to keep the server running and "re-initialize" the users connections in some way? If so, how?

 

[unixqueen]

Hi

I'm quite sure what you mean by 'blind' to the server.
Do you mean that the users are logged off and can't log back on again. Some applications can be configured to only allow a user to log on once. In this case, when a user is disconnected without logging off, the login still exists on the server and either manually killing the user's processes or downing the server would also kill the user.

Queenie

 

[penguin1]

What I mean by "blind" is that the users are shown to be logged off in our application we use, but in addition, the UNIX server cannot even be seen. We have tried doing a PING after our T-1 has gone down and the PING just times out. It is only after we totally bring down the server and bring it back up that our users can "see" the UNIX server and re-log into the application. I don't think it is that application causing this, it appears that we don't have something setup correctly on the UNIX box. Any ideas?

 

[crowe]

if the users on the local side are still ok after the T-1 goes down, I would think that the UNIX box is ok. Are there any routers that might need to be checked out?

 

[penguin1]

Well, we don't think it is a router problem because we have NT servers that we can PING through that same router just fine. We were told by an UNIX technical person that "UNIX blinds itself to a Windows '98 PC when that PC loses connection to the UNIX server." Is Windows '98 the problem? This doesn't sound right to me, but perhaps it is. At any rate, I would like to find some way to re-establish a connection without having to bring down the UNIX server.

 

[unixqueen]

Hi Penguin,

It sounds to me like unix networking is failing, as you suddenly can't ping the server, and a reboot fixes it.

Are the people local to the site able to continue working?
Are any messages written to syslog, dmesg?
Is anything written to the nettl log? - Check especially for duplicate IP's!
Do you have EMS configured?
Does this box have multiple network cards?, if so you might have an intermittant faulty network card or faulty cable to the router.

When this problem occurs you could run a traceroute from the console to the remote site and see where it stops.

Cheers
Queenie