DHCP reply with opt82

[Alan DeKok]

Alex Jaliashvili wrote:
> The problem was in using "old" attributes:
> DHCP-Agent-Circuit-Id := "%{request:DHCP-Agent-Circuit-Id}"
> DHCP-Agent-Remote-Id := "%{request:DHCP-Agent-Remote-Id}"

  Uh... no.  The dictionary.dhcp that is included with 2.1.10 has the
*same* definition for those attributes.

ATTRIBUTE	DHCP-Agent-Circuit-Id			1	octets
ATTRIBUTE	DHCP-Agent-Remote-Id			2	octets

ATTRIBUTE	DHCP-Relay-Circuit-Id			1	octets
ATTRIBUTE	DHCP-Relay-Remote-Id			2	octets

  They are in the DHCP-Relay-Agent-Information TLV.

  If you have ANYTHING ELSE for these attributes, then you are not using
the correct dictionary.

> Changed them to:
> DHCP-Relay-Circuit-Id := "%{request:DHCP-Relay-Circuit-Id}"
> DHCP-Relay-Remote-Id := "%{request:DHCP-Relay-Remote-Id}"
> and now it works without any problem.

  Which indicates that you're not using the correct dictionary.dhcp file.

> Debug still shows old attibute DHCP-Agent-Circuit-Id and doesn't show
> remote id at all:

  If the debug log shows DHCP-Agent-Circuit-Id, it's because you are
using the OLD dictionary.dhcp file.  Go fix that.

  The OLD file had "DHCP-Agent-*" listed LAST, which over-rode any
previous definitions.

  The only issue left in the version you have is that it looks like the
*first* attribute inside of option 82 is printed in debug mode, and the
others aren't printed.  But a "tcpdump" shows that all of them are in
the packet.

  I'll put a fix for the debug output for 2.1.10.

  Alan DeKok.