HP Procurve 2324 (J4818A) possible port issue

[Jamie Gorman]

We have a suite of 15 RM machines and 1 machine has intermittent issues connecting to our RM CC3 (Research Machines) network. The machine when it has the issue responds in the following way :-

When logging into a user account it will get half way thru the logon process and abort back to the logon screen. Its only when your repeatedly logon to the machine with the same user that it will logon after approx 4 attempts. Also if the machine is rebooted then it will normally logon fine after that.

By the way our external support have said thats its a possible switch issue to do with a port or to do with Spanning Tree and PortFast.

Thanks in advance

Regards

Jamie

[Matt Hobbs · Accepted solution]

The 2324 is considered an unmanaged switch and as such does not support spanning-tree. On the 2324 all you can really do is plug something in and walk away.

Try a different port that currently works fine for another user, see if the problem stays with the port or follows this PC.

[Richard Brodie_1]

The 2324 is an unmanaged device, so almost anything that involves some of the ports having a configuration problem (and others not) is out.

All that's really left is a bad port, and generally that doesn't happen unless the switch is old or cooked or both.

The downside of it being unmanaged, is that you can't look at the statistics on the switch to look at the error rates. You could try pinging the suspect machine remotely (with a count of 100, say) and seeing if there any losses. At least, you would know there was a network problem then.

What I would be inclined to do, though, is swap the machine over with one of the good ones, see if the symptoms follow the suspect PC, and then get back to your support people. Not that I'm prejudging the outcome of such a test, of course.

[Jamie Gorman]

Hi guys,

Thanks for your replies, and apologies in the delay in gettin back to you.

After running my own test by moving machines around and changing which ports goto a specific backbone cable, our support contract agreed that the issue was pointing towards the suspect machine.

So a hardware call was raised and completed (replacing the motherboard with intergrated network card) but the ping test that I had run previously showed the machine had the same issue. The test is detailed below:-

1) A ping command is set on the server to
continuously ping the affected
workstation (which replies fine)

2) Then the workstation had the same ping
command setup (but to the server) and it
goes to "request timed out"

3) At the same time as the ping is started
on the workstation, the server ping turns
into "request timed out" as well.

Just out of interest I ran the test on another machine that was unaffected... the results came back the same....

The one thing I don't understand is that it only noticably affects one machine...

Thanks for the replies in advance.

Jamie

[Richard Brodie_1]

Sounds like a paranoid firewall or IDS at the server end. If you have such, I'd definitely look at its logs.

[OLARU Dan]

Jamie,

congratulations for your currage of assigning points to the answers that almost got you to the solution :-)

If you are sure that any of the answers solved your problem, you should assign 10 points.

[Jamie Gorman]

Just to let you all know that a spare switch was installed as a replacement for the suspect one... And appears to be working fine.

Thanks for all your advice.

Regards

Jamie

[Jamie Gorman]

As discribed above the switch has been replaced and all is now fine. :-)