Unsolved

This post is more than 5 years old

8 Posts

13041

July 3rd, 2007 23:00

6224 STP causing packet loss at gateway

Hi,

Been having an issue with our 6224 which we finally narrowed down to STP re-converging and causing the switch to basically die for about 20 seconds.

We enabled fastlink on all ports to help with this and it has helped a great deal. But when we take down a port, then plug the port back it, it causes all the VLAN gateways on the switch to drop packets like made for about 20 seconds or so and we cannot do any management on the switch.

We did tests with Rapid STP, then MSTP, both with fastlink on every port (without fastlink it kills the VLAN gateways AND all traffic). With fastlink on it seems to just affect the gateways on all VLANs (not just the same VLAN the port is in). But traffic still seems to flow OK.

Now, this only happens with a VLAN that only has 1 port. On a VLAN that has 2 ports or more, if just one port goes down, then the VLAN stays up and nothing is affected.

Obviously I would like to find out why this is happening so I can have a port go down (ie. reboot a server connected to a port) and not have it take down the gateways or the entire network for 20 seconds when only 1 port is in a VLAN!

Running firmware: 1.0.2.6

Any ideas?

93 Posts

July 6th, 2007 17:00

Since your gateways are the VLAN interfaces, if you do not have any ports connected to a VLAN interface the VLAN interface will go down.  When you connect just one port to that VLAN interface it will come up and work.  If you do a show ip route you will see that without a port connected to the VLAN interface, that network will not be shown.

8 Posts

July 6th, 2007 18:00

I understand that the route will be out of the table when the interface and VLAN are down. But why does this scenario kill our switch? Basically, any of our customers can cause a denial of service on the switch just by rebooting their server (taking down the port briefly while server reboots). If the customer only has 1 port in their VLAN this is causing spanning tree to go crazy. I have several Cisco switches that don't show any noticeable latency or loss when routes, vlans, or ports go up/down. I am hoping this is just a configuration problem. Can anyone else offer any insight?
No Events found!

Top