Firmware 1.4.0.88 for SG300-52 appears to break DHCP Relay

[Jonathan Fisher]

My VLANs terminate at my switches, so I use DHCP relay with Option82 to hand out addresses for the VLAN subnet. After upgrading from 1.3.7.18 to 1.4.0.88, DHCP isn't being handed out on my VLANs. Anyone else having this problem?

[Jonathan Fisher]

Confirmed... I'm seeing the exact same behavior! Nice find.

Here are two packet captures from the SG300 to my DHCP server:

1.4: https://www.cloudshark.org/captures/f74bcfad752c

1.3: https://www.cloudshark.org/captures/76360f902726

[Aleksandra Dargiel]

Hi,

Yes, it does change but no issue with Windows 2008 server.

What is your DHCP server?

Regards,

Aleksandra

[Jonathan Fisher]

Dnsmasq version 2.68 

[Aleksandra Dargiel]

Hi,

Have you seen this working before with any other DHCP relay agents?

Aleksandra

[Jonathan Fisher]

Yes. dnsmasq is incredibly prolific, it's used in nearly every open source router. 

[Aleksandra Dargiel]

No doubt about this. Just there is a small difference between relay and regular DHCP discovery/request.

[Jonathan Fisher]

If you want to try to reproduce, Download a distribution that includes dnsmasq as it's DHCP server or just run Ubuntu from a live USB.

If there are other packet captures or anything else I can do on my network to give you information, I'm more than willing to help. I want to keep my firmware current but at this rate we'll be stuck on 1.3 forever.

[Aleksandra Dargiel]

Hi Jonathan,

I guess it would be good idea to open an official ticket with Small Business Support team so we will document everything in details:

http://www.cisco.com/c/en/us/support/web/tsd-cisco-small-business-support-center-contacts.html

Regards,

Aleksandra

[vignal-systems]

Hello All,

It's the same for me here. DHCP was working before upgrade but not after.

Windows clients can obtain IP on the vlan but wyses clients and windws CE clients fail to have IP via DHCP relay.

[Aleksandra Dargiel]

Hi,

Have not seen such an issue yet but if there is a problem with latest firmware I strongly recommend you to open ticket with Small Business team so they can narrow down and if needed open bug:

http://www.cisco.com/c/en/us/support/web/tsd-cisco-small-business-support-center-contacts.html

Regards,

Aleksandra

[vignal-systems]

I've no open contrat for our SMB switch so no support...

May be I've found a workaround for this issue. After many many try this issue seams to not be a DHCP relay issue but a global routing break that cause DHCP relay very very slow and cause timeout for some DHCP clients (in my case Wyse ThinOS).

The solution for me was to increase TCAM routing ressource from 128 to higher value.

We upgrade from firmware 1.0 to 1.4 and I belive that TCAM partitioning was quite different.

[vignal-systems]

That's why I could not reproduce the issue in test or mock-up pre-update systems.

Could you confirm that:

-Is it normal to have, for each host routed via SG300, 1 Neighbors TCAM entry? So Sg300 limit routing to -+400 hosts dependent of ACL / Services and routing interface in configuration.

-Do you have TCAM count of new Sg500?

Many thanks in advance.

[vignal-systems]

I reply to my previous post: as I can find on internet SG500X device is 3072 TCAM entries and the Sx500 device is 2048.

[Jonathan Fisher]

Is there a log entry or anything that would indicate it has ran out of TCAM resources?

[vignal-systems]

We have theses log entry :

Error   %ARP-E-ARPTBL: ARP Table Overflow, aggregated (3)       
Error   %ARP-E-ARPTBL: ARP Table Overflow     
Error   %ARP-E-ARPTBL: ARP Table Overflow, aggregated (9)       
Error   %ARP-E-ARPTBL: ARP Table Overflow