Re: squid-2.5.STABLE5 WCCPv2 Layer2 redirect

From: Henrik Nordstrom <hno@dont-contact.us>
Date: Thu, 1 Apr 2004 12:44:20 +0200 (CEST)

On Thu, 1 Apr 2004, Mark Foster wrote:

> Hi again, sorry for the delay in the reply.
>
> We haven't made a new parameter. At this time all the code is in wccpv2.c.

Ok.

> 1. squid registers with the router with forwarding method of L2 rather than
> GRE. This is the biggest benefit from this because it means that we don't
> have to muck about getting the packets back out of the GRE tunnel in the OS.

Good. A parameter for what encapsulation methods the cache server supports
would be appropriate here.. Not all routers supports L2 redirection, and
this also not wanted in all networks (the cache may be behind another
router)

> 2. we parse more of the I_SEE_YOU packet to cope with multiple caches. In
> particular we respond with ASSIGNMENT_REDIRECT with per cache bucket
> allocation.
>
> We have had this working with six caches joining and leaving the cachegroup
> and all works fine. We are about to let this loose on our user community
> using our group of four production caches so we'll get to see what it's
> like under load.
>
> There is a meeting today during which it will be clarified that the
> University does not have any issue with surrender of its intellectual
> property rights to the open source community; then we will roll our changes
> in to the latest available wccpv2 patch and submit a diff for your
> consideration.

Sounds great!

> Is this the latest patch?
> This patch is generated from the visolve_wccpv2-s2_5 branch of s2_5 in squid
> Fri Mar 19 23:20:33 2004 GMT
> See http://devel.squid-cache.org/

To my knowledge yes. There has not been any shared WCCPv2 development for
ages (years).

Regards
Henrik
Received on Thu Apr 01 2004 - 03:44:23 MST

This archive was generated by hypermail pre-2.1.9 : Thu Apr 29 2004 - 12:00:03 MDT