Recently we've installed several 3750 pairs in our server LAN. They are stacked with eachother and trunked back to our 6509's. Every so often the LED's on the active links flash amber, on older software revisions it used to increase the runt/input error counters whenever this happened, but on 12.1(19) its not anymore. I assume that's because Cisco considers the behavior to be a cosmetic problem and that it doesn't affect switch performance.
However, we were having problems with our SAP systems and they disappeared when we took the links back to the core out of trunking mode - which also stopped the LED from going amber and the error counters from increasing. Cisco wants me to put a sniffer on the problem, but the tech doesn't seem to understand that the switches are in our production environment that I cannot take down or otherwise cause problems on it. Its a 24x7 system.
All I know is if I have the link in 802.1Q trunking, I get the LED/errors and our SAP guys see the performance problem they were having. If I take it out of trunking, both go away.
Is anyone else seeing this problem? Do you know if the fix in the latest software that stopped the error counters from incrementing actually fixes the problem? I know the LED is still blinking amber quite often still - its just no longer incrementing the error counters... Seems like rather than fixing the problem they just hacked it to not show it as well.
What are the trunk modes on 3750 and 6509? Desirable? Trunk?
Try setting them both to Trunk so they don't have to negotiate the trunking and will always be trunks.
Also put a sniffer on the switch, it won't bring down the network, just span the traffic from uplinks to a sniffer port. Then you can capture the traffic and see if the frame header looses it's trunking characteristics.
I have been seeing the same problem too however I have found the following bug:
This is a cosmetic issue and does not affect performance.
CSCec14238 - 3750 connected using dot1q cause Runts increase on interface.
When using dot1q on trunk interface on the Cat3750, runts may be seen on show interface output because valid dot1q encapsulated packets which are 61-64 bytes including the q-tag are counted by the Cat3750 as undersized frames, even though these packets are forwarded correctly. In addition, these packets are not reported in the appropriate category (unicast,
multicast, broadcast) in receive statistics.
This is a h/w limitation of the Cat3750 platform. There is no workaround.
Yes, that is the bug that Cisco says that it is. I have tried changing the trunking mode, which we had set as desirable, to "on" and the problem seems to have disappeared on the switch we had upgraded to the latest IOS (12.1(19)). At my next window I will try this on the switches we initially had problems with to see if it helps with the problem or not.
Although Cisco says that the problem is completely cosmetic, if you are getting the LED flashing amber, from what we can tell it was causing degredation in network performance. Definitely upgrade and turn trunking on rather than have it set to desirable, this might fix the slight trouble that it was having.
[toc:faq]The ProblemOn traditional switches whenever we have a trunk
interface we use the VLAN tag to demultiplex the VLANs. The switch needs
to determine which MAC Address table to look in for a forwarding
decision. To do this we require the switch to do...
[toc:faq]Introduction:Netdr is a tool available on a RSP720, Sup720 or
Sup32 that allows one to capture packets on the RP or SP inband. The
netdr command can be used to capture both Tx and Rx packets in the
software switching path. This is not a substitut...
IntroductionOSPF, being a link-state protocol, allows for every router
in the network to know of every link and OSPF speaker in the entire
network. From this picture each router independently runs the Shortest
Path First (SPF) algorithm to determine the b...