atcomsystems.ca/forum Forums Computers & Programming Computers, Software, Accessories Network Problem

Hiya Folks,

this is my first time posting! One of the companies I provide computer support for has an office building with multiple offices inside including two large workrooms divided into cubicles.

One morning the receptionist went to log into the network but couldn't. The error showed 'network cable unplugged'.

I was called out and checked that her cable was plugged in to the jack which it was. The entire building was up/online and running except for the front 4 offices.

Each of the front 4 offices had the network cable unplugged error. Each office was wired straight back to the switch. All the other offices on the same switch were working fine.

The switches go to an 8 port router which had 4 empty ports.

These are the things I tried:

1)rebooting the server
2)rebooting the switch
3)moved the 4 ports on the router to 4 new ports
4)made sure there were no vlan's set up , they were all on the same ip network

I can see no central point of failure for those 4 offices out of the whole building since they are all wired straight back to the switch. The whole rest of the building was up and working fine!

I did not do the original wiring in this building and absolutely nothing has a label.

After 4 hours of being down the 4 office computers came back up online as mysteriously as they went down.

I just ordered a cable toner and am going to label everything. In the meantime trying to figure out why this happened is really bothering me.

If anyone has any ideas or suggestions , I would love to hear them.

thanks!

It seems to me you just about did all the right things concerning the hardware. That, and the fact that the connection reappeared several hours after all the hardware checks leads me to believe that some errand software may be the problem.
The message about the cable not being connected should not be taken literally. It's just one of the built-in error messages of the operating system that appears whenever a network connection is lost - for all intents and purposes, to the OS it looks as if might a cable was unplugged.
So the real problem becomes the loss of partial or full connectivity.
I would check the following:

1) If the PCs don't have static LAN IPs, check the DHCP server. It may have been down, or somehow misbehaving

2) Automatic Firewall/Security updates. Sometimes they can "break" part of the connection, then after the connections are "trained" (or after a subsequent auto-update) they can start working again. Some of these updates can affect the operation of DHCP as in #1 above.

3) Automatic Operating System updates. See above.

4) Any other software that may update automatically or uses the network. Sometimes, depending on the firewall configuration, repeated attempts to connect by software that's checking for updates or other external info are misconstrued as security risks, and the firewall may temporarily restrict access.

Good luck,
Perry

Hiya Perry,

thanks so much for responding!

I meant to put this in the network forum but must have clicked in the wrong one.

The DHCP server was up because the whole rest of the building was up and running.

You have given me some things to theorize about and check out.

I am going to take a picture of the wiring and post it in the thread with all the pictures.

thanks again!

You won't get a disconnect message if DHCP is down, you will get a limited connectivity message. You probably have a failure developing in the switch or your switch partitioned off the 4 ports. The reboot should have solved this although it could have been very temporary as in seconds. What brand and model is the switch?

You will need more than a tone and probe to diagnose network trouble. First you will need a cable analyzer - you can get good deals on the older model Pentascanners. You will need some kind of packet analyzer software and a portable laptop as well.

You don't want to be a road warrior tech and be swapping out hardware to see if it works. Fine for a car dealership mechanic but no good for you.

In this case the likely culprit is a switch issue. If the 4 computers were all on the same segment (4 is a magic number in most switches) then the switch shut down the segment that they were on most likely due to a bad network card at the other end. If the cvard was really cranking out garbage then the switch would have shut down that segment in a matter of seconds after reboot. Hard to know for sure without knowing the switch model and brand. Is it a managed or unmanaged switch?

What I would do is get a long patch cord and plug in to a office that is working to one of the pc that are not and you will be able to cut your trouble shooting time in half. If the pc starts working, look at the switch, if it still doesn't then you know it is an application, NIC, security settings.

A nice but older tool is the Compas by what was then Microtest. It could do a lot of quick diagnostics, wiremap, traffic analysis and link verification. They can be bought used for peanuts. You could carry a laptop with you that has all the same functionality today. Once in a while you get an IT guy on site who is so anal he won't let you plug in your own laptop to "his" network so those handheld analyzers are a good compromise.

Thanks for the responses!

mgere I did do that and the computer worked fine plugged into a different wall jack.

I am going to see if I can find some of that equipment on ebay.

PMCook thank you very much for that bit of information, I know that switches can fail or sometimes a port on a switch could fail

What I did not know was that 4 ports could fail at once. I kept thinking, what would be the odds of 4 failing at once. Your information changes everything and will be invaluable to me in the future.

I will post more switch details ASAP.

If it is a managed switch, get in there and look at the log file. Look especially for LACP errors. I have seen this before where LACP was negotiating on ports that were already assigned to host computers. When LACP failed the negotiation it shut off the physical layer which disconnected the device temporarily. The users were seeing disconnect messages at their end. If it is a managed switch and you are not using LACP turn it off.

binbrain, it would be helpful if you could post more details of the setup over there.
LACP errors are a possibility for a managed switch, but I noticed binbrain mentioned there were no vlans defined, which would likely mean that the host interfaces for the aggregation group would have to be explicitly set up. In that case, would LACP still try to use the already assigned ports?

Moved to Networking
click here