Take the 2-minute tour ×
Server Fault is a question and answer site for system and network administrators. It's 100% free, no registration required.

I am running a small network with 20 or so hosts and a machine running Ubuntu Server 12.04LTS as the gateway. Every few days I find that the network is frozen and that no two machines can communicate. The problem seems to be caused by one of two Linux PC's on the network locking up. I have been chasing the problem for a few months and here is a list of symptoms that I observe when the problem occurs:

  1. The activity lights for all connected ports on my switches are continuously flashing rapidly, even when there should be minimal network traffic.

  2. All ping requests between machines on the network give "Destination Host Unreachable" or "No route to host".

  3. One of the two Linux PC's on the network is powered on but unresponsive. Network activity light is flashing rapidly. Upon restarting, the syslog entries end abruptly without any indications of trouble. One is running Ubuntu 12.04, the other Mint 14.

Disconnecting the problematic machine instantly restores normal network activity, but the trouble machine must be hard reset to clear the problem.

Since the problem is tied to two machines with different hardware, but both have an Ubuntu based Linux distro I suspect a software problems. I have searched hard for a solution to this problem but am still quite stumped. Any suggestions?

share|improve this question
    
I take it moving the device to a different switch doesn't fix the problem? It seems like a case of a broadcast storm and could be software-related ...do the linux machines run anything in particular? –  Nathan C May 28 '13 at 18:41
    
Yep, it sounds like network flooding to me. What's on the Linux host in question? What kind of NIC configuration exists on this host? Does it have any NIC's in a bridged configuration? –  joeqwerty May 28 '13 at 18:59
    
The linux boxes are different hardware but both have the same NIC (Intel 82579LM onboard Gigabit). The Ubuntu machine is running 3.2.0-38-generic and the Mint machine has 3.5.0-17-generic kernel. No bridging or other interesting network configuration. The only software running is homebrew data acquisition and instrumentation control, but there is typically no software running when the problem occurs (usually overnight). –  Mike May 28 '13 at 19:42
    
Is there a bridge configured on the linux machine? If so, does disabling it fix the problem? –  Cian May 28 '13 at 19:43
    
I do occasionally run VirtualBox on one of the trouble machines (possibly in a bridged configuration) and I have noticed that it is more likely to crash when VB is left running overnight. –  Mike May 28 '13 at 19:49

1 Answer 1

It looks like you have some packet storm/loop. Ideally you should segment your network ;-) It could be faulty switch (oh! d-links few years ago!)

Wireshark is your friend here - run it fore few minutes and I bet you will be quick to spot what's causing an issue.

I would check: network mask, broadcast IPs, all multicats etc.

share|improve this answer
    
There are presently three switches, all TrendNET. One 16-port and two 8-port. I have tried switching out for D-link and Linksys switches with no effect on the problem. Power cycling the switch that the problem machine is connected to will usually fix the problem temporarily, even if the packets keep coming. –  Mike May 28 '13 at 19:45
    
netmask, broadcast, etc. are all set by DHCP on the problem machines. They seem to be set properly for my network. 192.168.0.255 and 255.255.255.0 –  Mike May 28 '13 at 19:48
2  
Wireshark/tcpdump. –  Chris May 28 '13 at 19:51
    
Wireshark has already uncovered several configuration problems with machines on my network, but nothing yet that relates to the machines of interest. I will have to wait for the network freeze up to recur to see what packets are flying around. –  Mike May 28 '13 at 22:36
    
I took some Wireshark data today when one of my machines crashed with this problem. The only activity seen by WS was the host machine looking for the rest of the network. As soon as the trouble PC was rebooted all normal network activity promptly appeared in WS. The log files on the problem machine show no problems only a gap in activity. –  Mike Aug 27 '13 at 20:22

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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