coincidence?

[bkaren1278]

For the last 2 weeks in a row i have reports of connectivity loss at ~540 PM. I have a syslog server that is not showing anything.. The loss seems confined to a 3500XL that has a fiber link to a 4507. All ports on the 3500 are on VLAN1. Some users have hubs/switches but they do not have spanning-tree portfast enabled on those ports.

the syslog shows a few interfaces going up and down but nothing out of the ordinary.

the switch's uptime is over 3 days

the IOS is at the latest 12.0(5) revision of 9/05.

what can be going on here?

[Georg Pauwen]

Hello,

it sounds like it could be related to spanning tree, especially since you say that some of your users are having hubs and switches. If the 4507 is your central switch, make sure that this switch is the root for all your VLANs. You can use the command:

spanning-tree vlan X priority 0

to ensure that. Can you try this and check if the problem persists ?

Regards,

GP

[bkaren1278]

would that explain that it happens once a week at the same time?

I thought disabling spanning-tree portfast allows switches to connect to the switchports.

[Georg Pauwen]

Hello,

first of all, I would be very careful with having users attaching switches to the network, this can lead to all sorts of problems, such as loops, or one of the switches becoming root for (a VLAN in) your network.

My thought was that your problem might be caused by somebody either connecting or disconnecting a switch, causing spanning tree to converge...

Regards,

GP

[bkaren1278]

makes sense, we disable portfast on all switchports going to switches.

i found this in my 4507:

2w1d: %C4K_CHASSIS-7-FANSGOOD: Fan failure fixed or new fan tray inserted

2w1d: %C4K_CHASSIS-2-INSUFFICIENTFANSDETECTED: Too few working fans, the chassis

will overheat. If not resolved, in 5 minutes all line cards will be placed into

Reset-Mode

i think this may explain things