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

I just built a server using a Supermicro X8DAH+-F board and running Ubuntu 10.04 Server 64bit. This has the Intel 82576 dual port controller (one port is disabled). Since this is a server, remote access is imperative.

The server is connected to a switch (DLink), and the switch is connected to a router running DD-WRT (Netgear WNR3500v2/U/L).

eth1      Link encap:Ethernet  HWaddr 00:25:90:03:c9:b9  
          inet addr:192.168.0.100  Bcast:192.168.0.255  Mask:255.255.255.0
          UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
          RX packets:7655 errors:0 dropped:0 overruns:0 frame:0
          TX packets:5772 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:1000 
          RX bytes:7179394 (7.1 MB)  TX bytes:919727 (919.7 KB)
          Memory:fbc60000-fbc80000 

lo        Link encap:Local Loopback  
          inet addr:127.0.0.1  Mask:255.0.0.0
          UP LOOPBACK RUNNING  MTU:16436  Metric:1
          RX packets:637 errors:0 dropped:0 overruns:0 frame:0
          TX packets:637 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0 
          RX bytes:96955 (96.9 KB)  TX bytes:96955 (96.9 KB)

I am pulling my hair out. This server randomly drops all connections. If I am logged in via SSH, the session will get disconnected between 0 mins (immediately) after login, to 30 mins. Once the connections are dropped, it takes several minutes for services to come back up.

I decided to run a 24 hour ping test from the server to the router. I have noticed that these disconnections occur during random periods of high packet loss between the NIC and the router.

The server is not overloaded with I/O processes or CPU processes and I am the only one using it.

Things I have tried to no avail.

  • Swapping cables
  • Swapping routers
  • Swapping ports on the router
  • Removing network-manager (Ubuntu)
  • disabling all firewalls
  • disabling iptables.
  • restarting all of the services manually.

I am considering buying a PCIe NIC, but I want to ask in case there is something I am overlooking.

share|improve this question
    
How about the eth0? Same thing or not? –  grs Jul 24 '11 at 21:33
    
There is no eth0 on this system, only eth1 and eth2. –  Ryan Rosario Jul 24 '11 at 22:54
    
My point actually is - does the other NIC suffers in the same way? –  grs Jul 24 '11 at 22:56
    
Yes, neither port works. Now, I cannot even get a response from DHCP. Starting to think this board is toast. –  Ryan Rosario Jul 25 '11 at 8:56
add comment

2 Answers

up vote 0 down vote accepted

One thing you might want to verify is that there are no other machine/device on the network "stealing" the server ip. Unless you can find that info in your network equipment there is always the option of running a arpwatch daemon on some suitable server on that local network.

share|improve this answer
    
This was the case. The old machine had a static IP address assigned by the router based on its MAC. Although I changed the MAC in the router, rebooted etc. etc. the old server would not give up the IP. Even with it powered down, the IP would not take the IP away. –  Ryan Rosario Jul 25 '11 at 20:53
add comment

Since both on-board NICs are losing packets and you did many tests, the only one left is to try separate standalone card as you think about it. This will answer your questions.

share|improve this answer
add comment

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.