0

We are having an intermittent issue with a campus of Alcatel-Lucent Omniswitch switches, ranging from model 6000 to 6850. We are having a problem tracking down the root cause of our issue. The seems to be some rogue traffic or something causing all of our switches to slow to a crawl. Connecting to the console port on any switch and issuing the CLI command "show health" shows us that the CPU load is pegged at 100.

There are many VLANs on the switches, however, whatever traffic that is causing this, is traversing all the switches within seconds. OmniVista (a monitoring program from Alcatel-Lucent) reports a whole bunch of new spanning tree activity happening as the switch CPU load goes to 100. Our network scheme is a star, with one core stack switch and all the other switch stacks hang from a port on the core. There are a few secondary switches with a port connecting a third level switch. There are no physical possibility of someone, outside of the server room, to run a cable from one star-branch switch to another to create a network loop.

We've been on the phone with Alcatel-Lucent support for hours at a time for almost a month now, without any solution.

Can a single computer or printer cause this kind of enterprise-wide failure across switches? Can it be stopped or prevented?

2
  • Is there any layer3 work being done on these switches?
    – pauska
    Mar 1, 2014 at 14:57
  • None that I am aware of. I inherited this network from someone. How do I tell if there is any Layer-3 stuff going on?
    – J. Chin
    Mar 1, 2014 at 19:24

1 Answer 1

1

An easy way to check for L3-configuration on a OmniSwitch is to issue the command 'show ip interface'. If there are more than Loopback and another interface for management over ip you can assume that routing is being done.

'show configuration snapshot' will print the whole configuration and that will be of help while debugging.

My first tip would be to ask yourself if you really need Spanning Tree. I've seen Spanning Tree misbehave, most likely because it is hard to manage.

Also, because you are seeing a lot of Spanning Tree topology changes, you might be dealing with loops or rogue BPDU-frames.

The topology changes may also be caused by switches not being able to send BPDU-frames at required intervals which causes other switches to set "dangerous" ports to a blocking state.

Try to pinpoint which switch the other switches start to block, the issue may originate from that one.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .