N2000 - arp issues with Enhanced hashing mode

Question

Hello, I would like to report a problem I had after upgrading a stack of three N2048 switches from version 6.2.6.6 to version 6.3.0.16 After the upgrade, some PCs in some vlans could not reach their gateway anymore. I found that these devices didn't have their gateway's mac in their arp table. It would seem that their arp requests would get dropped in the port channel connecting the N2048 stack to the rest of the network. Adding a static arp entry on the affected PCs allowed them to reach their gateway. I know that there has been a bug (fixed in 6.3.15) with packet drops on port channels using mode 7 (Enhanced). This gave me a clue and today I changed the hashing mode to 6 (Source/Destination IP and source/destination TCP/UDP port) and I removed the static arp entries I had created as a workaround the day before. I could then see the gateway as dynamic entries on the PCs, as it should. It would seem that there is still a bug with Enhanced Hashing in 6.3.0.16. I'll be happy to send my config and / or more information if requested to help you investigate the problem. Regards, Martin

jameshead79 · Accepted Answer

cool, that's literally just been released although there have been a lot of releases in recent months. Cheers Jamie

FlatB · Answer

I can confirm similar issues to report on this. Moved to 6.3.0.16 and this issue started.

txag1995 · Answer

Thanks Martin!  I began noticing this issue today.  We're on 6.3.0.15, and switching to mode 6 got us back in business.

jameshead79 · Answer

Hi All,

Yes, I had the misfortune of encountering this issue after I upgraded my 2048P stack to version 6.3.0.18.

This will impact a switch stack that uses a port channel split across different switches.

I was using a stacking group of 6 switches. The port channel was connect to switch stack member 1 and 6. Hosts connected to switch 1 and switch 6 couldn't break out of their own default gateway.

I also had all my wifi AP's connected to the offending switch stack members and all my wifi users also experienced connectivity issues.

As mentioned in previous posts, changing the port channel hash algorithm to 6 resolved the fault.

Shame this wasn't highlighted as a warning in the release notes.

Regards,

Jamie

Networking General

Was this post helpful?