Why would an entire building have 169 IPs but rest of the site is fine?

[TomHD]

Right, thanks for the responses guys. I have isolated some non working Ethernet ports and got the building back up and running as it should. I still worry there is an issue and the problem might crop up again. As its working and its Saturday I'm going home Fingers crossed all is well on Monday!

 

Thanks again to all those involved

 

Best and hoping all have a good weekend. Tom.

[m25man]

I have posted this elsewhere but it may be relevant.

Apart from the "possibility" of the infamous malicious loop cable which is my own experience rare, there are dozens of other possibilities to cause this.

 

Firstly do a packet capture using an analyser, wireshark will do to identify the source of the broadcast or multicast.

 

The Multicasting issue has been around for sometime but in recent weeks probably related to many new roll outs or system image deployments PC's with Intel LAN cards can begin to multicast when in standby.

This is intermittent and cause by a bug in the hardware and certain drivers.

If your switches are not set to handle Multicast Packets you can loose complete sections of your LAN.

Make sure you have IGMP Snooping configured correctly. Update any machine using Intel Lan cards to the latest drivers do not use the OEM or Windows own drivers.

We arrived at a site with a similar condition during the break, they couldn't login or get an IP address.

They were the only machine actually on at the time but the switches in this building section were going ape!

A quick check on the Netgear switch web management identified 3 ports generating millions of multicast packets.

The offending machines were brand new HP's that had been deployed using the same WDS image and they were all in standby!

 

As soon as they were woken up or the switch reset the multicasting stopped.

Updated the driver including the one in the image being deployed and the issue stopped.

 

Other causes include, bad cabling, poor punch downs, twisted/damaged patch leads.

These can easily cause excessive broadcast and switch issues if a single pair becomes open/short/intermittent.

All 8 wires are needed for gigabit if one is broken or short the machines can suddenly drop out to 10/100 or loose the RX link completely effectively becoming deaf temporarily.

They begin to broadcast frantically and as they are deaf don't know when to shut up!

Once again better quality switches with broadcast storm controls can minimise these types of issues.

 

Just because a switch is cheap is not enough to condemn the entire brand, especially as we don't know if the rest of your LAN is in good health!

(Although I too would tend to avoid using Planet Switches on a Campus LAN).

 

I was in a brand new building in Docklands yesterday, there was an argument going on between a BT engineer and a cabling contractor because an emergency phone wasn't working in a lift.

Mr BT came into the Comms room and plugged in his handset, got dial tone. Went to lift no dialtone!

Cable guy said it can't be his cabling as they were all tested and certified!

But the Building Inspector guy said - No emergency phone no building opening - that was it.

 

With nobody else equipped I offered the cable guy a terminator from my Fluke Test kit and asked him to plug it in the lift shaft RJ45 port.

I immediately got a Pin 1 Or/Wh open circuit at his end.

There you go I said, you will need to re-punch that outlet...

Sod that he said, just move the patch cord to the next port in the patch panel.. hey presto it all worked!

 

When he came to return my terminator I said what about the one wire that needs re-terminating? Not my problem he said...

I counted well over 1000 RJ45 outlets there, how many more will there be in that brand new building incomplete?

 

You will never be able to trust your cabling until you have verified each outlet yourself, do not write off the switchgear until you are certain your cabling is good.