Plugging in a wireless router crashes a section of the network, randomly

[Kaelstorm]

I haven't been able to locate any information on a problem like this anywhere else on the internet, but maybe someone has an solution for what I'm describing. I'm an Intern at a business that currently has no WiFi. Back in December the company network was down for about 2 days until they unplugged all the wireless routers, and just now we're working on restoring wireless. We've purchased brand new routers, and have been testing just one at a time to see if there's any problems with the overall network. Usually there is no problems, but randomly (anywhere from 2 days to 5 hours) two people will lose their internet connection. Their VOIP phones go offline and they have no connectivity. It seems to be the same two every time, but there's never any noticeable reason as to why this happens.

The routers are configured for the LAN network, mainly to function as access points with statically assigned IP's that fit within the reservation list of the DC, and currently only 1 is plugged in at a time to test.

If anyone has any ideas as to why a wireless router (with no one connected wirelessly) could crash two people on the network please give me some advice. If anyone needs more information let me know.

 

[kanewolf]

Conflicting DHCP servers would be the first thought. Ensure that the DHCP server on all these routers is disabled. Next would be static IP conflicting with DHCP address range.

If you had to buy devices, why did you buy routers instead of dedicated access points?

 

[Kaelstorm]

Conflicting DHCP servers would be the first thought. Ensure that the DHCP server on all these routers is disabled. Next would be static IP conflicting with DHCP address range.

If you had to buy devices, why did you buy routers instead of dedicated access points?

The sysadmin wanted new routers, because that's what had been used throughout the building in the years previously with no issues, and he thought maybe it was a simple hardware issue causing the problems. These issues just arose this past December as far as I'm aware.
The DHCP is definitely off on the wireless routers, and the LAN IP's that the routers are set to are reserved by the DC.

As far as I'm aware, (I've only been here three weeks), all the networking hits a layer 2 switch, then to the cisco router, then to the outside modem.
To do some testing with seeing what causes the network outages, we ran a line straight from the Modem to an un-managed 8 port switch that the wireless routers was connected to, to give his office access straight to the outside and had no issues for the short time window that we tested it.

We tried another cabling configuration where we bypassed the layer two switch, by going straight to the cisco router from the unmanaged switch that the wireless router is attached to, and within 30 seconds the entire internal network within the building was crashing to varying degrees of severeness. Some people still had connectivity others had nothing.

When the wireless router is plugged into the network, so that it goes through the layer two switch, and then to the cisco router, and then out to the internet, there usually are no issues from upwards for 30mins - 3 days. But at some point the network will start crashing.

My thoughts is that it's some issue that developed with the DC, but the problem was here for a month or two before I started.

Sorry if this explanation is confusing, my networking skills need improvement.

 

[kanewolf]

You probably need that core switch to be a managed switch if it is not already one. Basic networking, what can go wrong? Route? DNS? DHCP? Loop? Packet Storm? You are going to have to look at the basic things. Maybe setup wireshark to see what is happening. Maybe work on forcing the error to happen during off-hours when you can be more invasive in troubleshooting.

 

[Kaelstorm]

I asked my SysAdmin to take a look at the switch. It is a managed switch. We were looking at the packet information on the ports, and one port (port 8 in this case) had like millions of failures, packets collisions, dropped packets, etc, while all the other ports were fine.

Does this sound like a likely culprit? Because of the way the building is wired, it used to be a large old factory converted into office space over the last 20 years, figuring out what machine or unmanaged switch was plugged into that port is going to take a long time to identify, because the Ethernet cable just disappeared up into the ceiling and then through a concrete wall out of sight, and walking around looking for things offline hasn't been effective. Also no ones come running saying their network is down.

*Never mind, we found out where the network went down, and it really wouldn't explain the wireless router issue. But I am curios as to why the was like a million errors on that port."

 

[kanewolf]

The managed switch will have a MAC to port table that you can use to help identify what is on a port. Millions of packet errors could be a problem. The switch could see it as a DDOS attack and do a reset or shut off ports or something.

 

[bill001g]

It depends on your equipment. All you would have to do is plug a switch/router into that port and then plug 2 ports together with a cable. You get a instant loop and broadcast packets will eventually overwhelm that segment. At that point you generally see all kinds of different errors on a port.

Spanning tree and other features are used to prevent that. What I was surprised to find is many $10 unmanaged consumer switches have that feature enabled so it is much less a problem than it used to be.

Can't tell you how many times we had people be bored in a conference room and hook two jacks together on a table. Luckily we always ran spanning tree so it just caused a alert message and blocked the ports.