On Monday 07 January 2002 08.04, Henrik Nordstrom wrote:
> Where Cisco found their protocol type 883E I do not know, but it is
> certainly not mentioned in any of the above documents, not even listed
> as reserved by Cisco. So any GRE implementation supporting WCCP can in
> effect be argued as broken.
>
> But sure, having (optional) support for such brokenness can indeed be
> nice.
Further investigation clearly indicates the support needs to be optional
and not blindly enabled. Cisco redefined the meaning of their GRE Protocol
883E in WCCPv2 to have a quite different meaning than WCCPv1 (now there is
suddenly a kind of WCCP protocol header of fixed size, with no change at
all at the GRE layer). So without actually knowing the endpoint sending
you packets of GRE Protocol 883E you cannot know for sure the format of
these packets short of guessing based on size and magic numbers in the
real payload.
And there is no version number in the WCCP packet header, so next change
of the meaning of the fields will totally confuse things again.
Regards
Henrik
-- MARA Systems AB, Giving you basic free Squid support Customized solutions, packaged solutions and priority support available on requestReceived on Mon Jan 07 2002 - 02:53:53 MST
This archive was generated by hypermail pre-2.1.9 : Tue Dec 09 2003 - 17:05:39 MST