2510-24 (J9019B) acting odd when broadcast traffic is high

[John H. Miller]

I have 27 of the subject switches, all running Q.11.67 software and all exhibiting the same issue. We recently had a computer on the network go crazy and was broadcasting ARP traffic at an alarming rate. This did not seem to cause any real issues, except that all 27 switches had their PING response times shoot up, and I mean way up. They were all averaging around 800ms. Throughput did not seem to suffer, but the high ping response time did cause network monitoring issues (using Cacti which relies on PING and SNMP). Additionally, the CLI response time was slow. Once the source of the traffic was removed, the response time returned to normal.

None of the other switches on our network exhibited this, only the 2510-24's, and it was all of the 2510-24's.

Could there be a firmware issue with this switcch that causes the problem? Has anyone else ever seen this phenomena? Agree that hi broadcast is not an ideal situation, but should it sut donw some of the switch functionality?

Thanks folks. I really value this forum and look forward to what you might provide. -  John

[Vince-Whirlwind]

The switches aren't the problem. All layer-2 switches should put responding to pings very low on the list of priorities, so when CPU jumps up, it will be slow creating ICMP packets. This is normal behaviour. Your Cacti went crazy because you had a problem on the network, not because the switches were doing anything wrong.

The actual problem is your ARP storm - you could mitigate against this sort of thing happening again by implementing a broadcast limit on the uplink port for each Access-layer switch.