Announcement

Collapse
No announcement yet.

Fiber loop packet loss

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Fiber loop packet loss

    We recently brought another remote location online. We created a fiber loop when this building went up. Now, when we complete the loop, it creates packet loss and devices intermittently lose connection. It's as if the devices are unsure which direction to send the packets. Here is a diagram of how its setup now--



    Right now the connection between Switch 4 and 5 is unplugged. This is not good because if Switch 2 goes down than the new building at switch 3 would be offline. Switch 3 (the new one is a Cisco 3750 and connected with single-mode fiber to switches 2 (3550) and 4 (3560 8-port w/SFP). I feel like I am overlooking something crucial but just can't put my finger on it right now. Any ideas?

  • #2
    Re: Fiber loop packet loss

    A loop in your network is never a good idea.
    How are the devices connected?

    IIRC you have to use Spanning Tree
    http://www.cisco.com/en/US/docs/swit...ide/swstp.html
    Marcel
    Technical Consultant
    Netherlands
    http://www.phetios.com
    http://blog.nessus.nl

    MCITP(EA, SA), MCSA/E 2003:Security, CCNA, SNAF, DCUCI, CCSA/E/E+ (R60), VCP4/5, NCDA, NCIE - SAN, NCIE - BR, EMCPE
    "No matter how secure, there is always the human factor."

    "Enjoy life today, tomorrow may never come."
    "If you're going through hell, keep going. ~Winston Churchill"

    Comment


    • #3
      Re: Fiber loop packet loss

      Are you running spanning tree on these devices? If not spanning-tree will take care of the loops and put some links in a blocking state. You need to look at how your switched network is designed. Do you have a root bridge defined? Are there redundant links in the topology? One switch going down should not cripple your network.
      CCNA, CCNA-Security, CCNP
      CCIE Security (In Progress)

      Comment


      • #4
        Re: Fiber loop packet loss

        Spanning tree is enabled using just the default settings. When switch 4 and 5 are connected, devices that are connected to switch 5 intermittently go up and down. We ran a ping -t from some workstations on switch 5 to the gateway and about every 10 pings it will timeout about 5-10 times and then come back up and then do the same. The root for the vlan we are having issues on is actually a switch connected to number 6 in the diagram. I am thinking of manually changing the root to switch 1 as that is our main core switch (Cisco 4510R) and also the root switch for our native vlan (vlan 1).

        Comment


        • #5
          Re: Fiber loop packet loss

          Switch 5 is prolly having issues when you connect switch 4 because spanning-tree probably detects a loop in the topology which is what it should be doing. If you run spanning tree some links will be blocked its unavoidable. Why not disconnect switch 2 from switch 3. That should let you connect 4 to 5. Depending on cost etc the traffic flow if you change sw1 to the root will be

          Switch1 (root) - SW6 - SW5 - SW4 - SW3
          Switch1 (root) direct connection to SW2

          If your worried about single point of failure run multiple links with etherchannel. Spanning tree treats multiple links in a channel group as 1 logical link and will offer redundancy if a link fails.

          You can also connect SW2 and SW3 directlly to the root SW1 and keep the existing topology which should work.

          This is not good because if Switch 2 goes down than the new building at switch 3 would be offline

          This shouldnt happen if you follow the above. SW3 will have a route back to the root via 4-5-6-1. This is assuming you make SW1 the root. Once you define the root make sure there is no way for another switch to take over as the root. Root election is based on Bridge ID and Mac address. Bridge ID is 32768 by default. If there is a tie the mac address will be the tie breaker. Older switches tend to have lower macs then new switches so its very possible when you add an older switch into the topology for it to become root. Thats why its important to statically assign your root bridge.

          Also check for bad fiber cables. Its possible with a bad fiber cable to have one way communication, which means you can send BPDU's but the other side may not be able to process them or send BPDU's of its own. Typically if a port doesnt hear BPDU's for maxage the port will try to transition to a forwarding state which can also cause a loop. Cisco usually has UDLD enabled on fiber ports by default to combat this. I would look and see what you spanning tree path is right now. Which ports are forwarding and which are blocking. Run some show commands and maybe even some debugs.
          CCNA, CCNA-Security, CCNP
          CCIE Security (In Progress)

          Comment

          Working...
          X