2924XL's 'pausing'

[acscco]

We have 60+ 2924XLs and a handful of 2912s. We had an issue last year, where they would randomly stop passing traffic for several minutes at a time. We never figured out what the problem was, but all of them seemed fine after replacing them with replacements from Cisco. So, we assumed that it was a hardware issue. After 1-2 months of working well, they started doing this again. It is getting really bad now, and is causing some headaches for us all.

1) 99% of them are running 12.05 (WC11) code

2) updating the code to WC14 does not seem to help (I haven't tried WC15 yet, but will now.)

3) most are trunked into 3550s or 3560s

4) all trunk links are manually set to 100/full

Oddly enough, when I console into one that is down, it immediately comes back up. I can also telnet to its uplink switch and ping the downed switch. This brings it back up.

Thanks for any ideas that you might have.

Alan Surrette

[Edison Ortiz]

How is your VTP configuration ? I hope you aren't running all of them in server mode.

[acscco]

I just went through them (all of our switches) back in March to make sure that all of the VTP and SNMP settings were correct. I can do that again, if it sounds like that might be the issue. They should all be in client mode except for the 6509 at the core.

I am getting ready to swap out some switches with brand new 3560Gs for testing purposes.

Here are some more things that I have found today -

1) There are at least 2 2950s acting the same way.

2) Once I get the switches up and running, a continuous ping to the previously dead switch will keep it up. As soon as I stop pinging it, it goes down again. Then, I have to either wait, console into it, or ping it from its upstream switch to get it back up again.

I am going to replace one of the flaky switches at one site, and one rack core switch at another site. I'm thinking that maybe the cores are getting overloaded. (I really just want to rule that out.)

Checking the VTP settings again is not a bad idea. Thanks!

Alan

[acscco]

I just notice that sometimes the switch is not passing any traffic. There are other times when it just does not respond to pings or allow telnet. The one that I am looking at right now was un-pingalbe until I connect to the console and hit 'ENTER'. Then it started responding. However, I am also connected to one of the ethernet ports, and it has worked fine the whole time that the switch was unresponsive.

I have seen them not pass traffic though...

Alan

[Edison Ortiz · Accepted solution]

In addition to VTP, verify STP configuration all around.

[acscco]

It appears that the STP configuration was the major problem. THANKS! We had added some VLANs, but had not configured the spanning-tree priority for those VLANs on the core switch. There appears to be some 2924XLs that do still have hardware issues, but the network-wide problems have calmed down considerably.