3850 UDLD Error after upgrade to 16.6.7

[Joris Deprouw]

Hi Guys,

I have a strange issue with several WS-3850-24XS switches.

Recently we have upgraded 14 WS-3850-24XS switches from version 16.3.7 to 16.6.7.

After the upgrade of the switches we noticed on 5 out of 14 devices that several interfaces (between 1 and 10 ports) were in err-disable due to UDLD.

We reloaded the switch a second time….without result. When tried replacing the GBIC, but this did not solve the issue.

When resetting the old uplink port to default and re-applying the config, the issue is not solved.

When moving the existing GBIC to another free port and re-applying the same config the issue is solved.

We have moved all the impacted ports to other free ports on the same switch, so everything is working fine now.

I have upgraded several customers to version 16.6.7 without issues. I’m wondering if it’s maybe hardware related. Since only 5 out of 14 switches are showing this issue at this customer.

Thanks,

J.

[Georg Pauwen]

Hello, 

I checked for bugs, none really seem to apply. Does this happen in both udld and udld aggressive mode ?

[Joris Deprouw]

Hi Georg,
I didn't try to change the UDLD mode, but in this case UDLD normal mode has been activated on all of the switches. Their config is based on a template, so UDLD normal mode is the standard setup, also configured on the switches which didn't have the issue.
I didn't find any bugs either...16.6.7 should be a stable version according to Cisco tac time.
Br,
Joris

[Mark Malone]

Hi
Ive hit issues like that on 16.6.6 , we did same thing moved to another 10GB uplink in the 8 port 10GB module , i dont think its hardware , just wasnt bothered upgrading as its stable image other than that but ye ive seen that problem between floors in remote offices where we are running 16.6.6 on 3650s , same ios-xe as 3850

oh an we especially seen it in stacks more , the second module in second switch first couple ports wouldnt work for a while kept going err-disabled but we eventually got them up after a few reboots

[Joris Deprouw]

Hi Mark,
Thanks for your feedback. We actually don't have any stacks. So they are all standalone switches.
So it might be an undocumented bug. I might try to upgrade to 16.9.4
Thanks,
Joris

[Georg Pauwen]

Hello Joris,

the only thing I could find was something about a missing native Vlan in the Vlan database that might cause the errors. Do you have trunks configured with a different native Vlan than the default (1) ?

[Joris Deprouw]

Hi Georg,
Indeed I saw that same issue as well. Not different native vlan in our configuration...
Br,
Joris

[Leo Laohoo]

What ports are these? Where they 1- or 10 Gbps ports?
If these are 10 Gbps ports then look at CSCvo55360 &/or CSCvg48154.

[Joris Deprouw]

HI Leo,
The ports are 1G and 100M.
Br,
Joris

[Leo Laohoo]

What? 100 Mbps? What module are these issues appearing on, GLC-TX????

[Joris Deprouw]

The impacted ports have a mix of (GLC-SX-MM) 1000BaseSX SFP and (GLC-FE100FX) 100BaseFX SFP inserted