Incorrect time settings on switch causing port disconnects?

lynne.meeks · ‎03-24-2008

Has anyone ever seen an incorrect date/time setting on a switch cause ports to randomly disconnect/reconnect?

We had a report of connections being dropped during on on-line exam; each 48 port 3560 gig switch was 80% active connections, and we would see about 6-8 ports go down and up each minute.

There were no errors on the switch or any of the ports.

However, I did note that the date stamp was incorrect; it was off by a couple of weeks (early). I corrected the date stamp on the switch so the logs would be correct --- and the random disconnects ceased! This behavior repeated on 3 different switches.

I have never seen an incorrect date/time stamp cause this kind of issue, and I would love to to see an explanation.

thanks-

Lynne

glen.grant · ‎03-24-2008

That would not have anything to do with it , the problem is elsewhere. Did you check for any code bugs in the version you are running?

Richard Burts · ‎03-24-2008

Lynne

I wonder if there is something in your environment that might cause this behavior. I note part of your description is: connections being dropped during on on-line exam. I wonder what is going on with on-line exam?

Is it possible that the end stations connected to these ports are hearing/observing/learning time (via NTP, or SNTP, or Windows time service or something) from the switch? And is it possible that if the time they learn differs significantly from their internal time that they might reboot as part of "fixing" the problem that they observe? A reboot from the end stations would be one thing that could make the switch ports go up and down.

I agree with Glen that an incorrect time would not make the switch to change ports up and down (or down and up). But I am not sure that we know whether it is a switch issue or an end station issue.

HTH

Rick

HTH

Rick

lynne.meeks · ‎03-24-2008

I will look for bugs- but there is a completely different version of code on the 1 3750 than there is on the 2 3560s- I will look however.

I agree that this makes no sense, but the change was so instantaneous and dramatic and on 3 separate switches that it seemed it must be related.

I have been trying to think of what possible correlation there could be between the switch time and the end-user status as you suggest Rick, but I havent' yet come up with anything.

We don't do any NAC, and I do not believe the end stations rebooted; they just lost their connections.

_twilight zone music plays here...

Lynne