19

I was searching for a LAN adapter on Amazon.

One of the reviews stated that this LAN adapter caused a crash of their home network. The router would restart a lot and the laptop wouldn't have an internet connection. Removing the adapter solved the problem.

In case you want to read the review, here it is (on German): https://www.amazon.de/gp/customer-reviews/R3P59KQU3K8MB2/

I am just wondering, how or if that could even be possible? Doesn't the LAN adapter just "convert" the signal and not mess with IPs? Shouldn't the router just not deliver the signal to a faulty LAN adapter instead of restarting itself?

Maybe the buyer had a bad router?

7
  • 1
    Is there a reason you'd want a USB device? They tend to require more CPU time to service compared to a PCI or PCIe NIC. If there are expansion slots in your device, then do consider using them. Not much help on a modern laptop sadly. – Criggie Mar 15 at 7:46
  • Note that as shown in the answers below, it's possible, but without more details about the exact setup and what happened exactly, it is quite difficult to assign the blame in this specific situation to either the LAN adapter, the computer it was connected to (and any applications — or malware — it was running, or its network configuration), the router, cabling, switches or anything else in the vicinity. – jcaron Mar 15 at 8:59
  • I think the key thing to know, when looking at the answers, is that these devices are connected by pieces of copper, and are well behaved because all of the devices agree what voltages to apply to the copper, and when to do so. When a device becomes ill-behaved, it can violate these "social contracts" between the devices and cause all sorts of problems. Typically it won't cause lasting damage, but it can stop things up pretty good. – Cort Ammon Mar 15 at 15:24
  • 1
    The reason for the USB device: I need an adapter for the Nintendo Wii (yes I am still using one) and the Wii can only have adapters with a specific chipset. But better safe than sorry, I buy another adapter now. – User123456789 Mar 15 at 20:39
  • If that adapter sends out a payload which gains root access to your router and deletes the entire filesystem, then yes. – john doe Mar 17 at 17:38
63

Yes, it's possible for an Ethernet device to sabotage the entire network. At work, a core switch died and started flooding the network with invalid Ethernet frames at ~700 MBits/s. This caused all PCs on the network to overload the CPU so much even moving the mouse no longer properly worked.

Of course, non-PC devices could be affected by problems like that as well.

13
  • 11
    If you have a bittorrent client running on a really slim router, with no regard for network capacity, the number of connections and packets can easily overwhelm the router and cause it to crash. This is a legit, functioning software, so I can imagine a bad device to do even worse. – Nelson Mar 15 at 1:57
  • 10
    It gets worse. An ethernet card could theoretically fail in a way that puts too much voltage on the pins. Short to mains should blow a fuse, but short to +5 will be a surprise for the home router. – Joshua Mar 15 at 4:01
  • 6
    Can confirm this answer as well as @Joshua .. once had a problem with a server that, when booted/rebooted, would cause an entire network segment outage because its onboard network card was shorted out on an IC level .. this was a dev server that was constantly rebooted (or crashed due to bug). Took nearly 3 weeks to make sure it wasn't a software problem (using good ol' oscilloscope) and eventually had to put in a PCI network card to solve the problem. Fun times that was. – txtechhelp Mar 15 at 6:13
  • 5
    @Joshua: IEEE requires Ethernet to reject 1500V @50/60 Hz for safety reasons. Short to mains may trigger a GFCI on the mains side but the router ought to be fine. 5V DC shouldn't be an issue at all. Ethernet was not designed as consumer-level, cheapest-thing-that-barely-works electronics. – MSalters Mar 15 at 10:19
  • 4
    @AndrewHenle: Lightning bolts are 300 million volts. 1500 Volt is a perfectly manageable voltage. Old TV's generally used ten times higher voltages without destroying themselves. High-speed trains have 25000 Volt overhead wires, and those are hanging outside in the rain. – MSalters Mar 15 at 12:17
23

Absolutely yes.

There are many ways an Ethernet device can kill a network, especially a network with low consumer-grade equipment.

It can spam out malformed packets, swamping the router.
It can have a MAC address that matches an existing device on the network. Yes, this should not be possible. Also YES, it can and does happen.)
It can have a null MAC address, and your network has a router or device that cannot work with this.
I have even encountered an Ethernet device that output 220V on a PoE line, toasting everything downstream!

If your network is managed by a real device, say a decent Cisco router that is correctly configured, it will simply detect and isolate the malfunctioning/misbehaving device. But at household level? Nope.

3
  • 1
    Meh, I've seen a whole company network taken down by one person who short circuited a dumb hub between two switch ports. Not the fault of the hub, mind you, but even enterprise network equipment can't deal with everything. – J... Mar 15 at 13:19
  • 4
    @J... well it is supposed to be network not a shrubbery – PcMan Mar 15 at 14:01
  • I've seen a network taken down by someone confusing cables and connecting two ports of the same switch. The loop multiplied broadcast packets and slowly but surely overloaded most switches in the network. To be fair, the administrator had STP and related features disabled on all switches because the periodic packets annoyed him, but yes even something like this can bring a whole network down and it can take a while to find the cause. – cg909 Mar 17 at 15:07
5

Possible, but quite unlikely.

On the other hand, I had a somewhat similar experience with rather poor quality router.

The router ran more or less normally over wifi (B/G), but plugging any 100MBps crashed the router at the first bulk download. The router simply overheated when pushed to its limits at 100MBps and ran more or less normally at the 18MBps that are the maximum for G-type wifi.

Forcing the connection to 10MBps made the connection stable until the router was replaced.

1
  • Additionally, poor cabling/patches/termination can cause issues too. As do cabling loops and occasionally hosts with two interfaces on the same network (wired and wireless even) – Criggie Mar 15 at 7:44
3

Yes, some combination of OS drivers and hardware chips cause issues, where the network adapter emits a network flood when the computer is put into sleep mode.

An example of this is: https://www.isc.upenn.edu/how-to/intel-i2xx825xx-network-card-network-flooding-issue-fix

In the above issue, the network adapter floods the network with IPv6 multicast traffic, and switches not supporting multicast will cause this to flood the whole network.

If a well known brand makes this mistake, what prevents a lesser known brand from making the same mistake?

0

If you want to try it for yourself (you probably don't), there is a very simple recipe: Attach a patch cable as a loop between two free ports (there are typically 4 switched LAN ports on the back of a home router).

The first time a packet for a non-local device appears, it will start looping between the ports at max speed and get replicated to the other ethernet ports as well, resulting in a devastating packet storm on all ports.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.