strange behaviour of WCCP

[maxxo]

I am implementing a WCCP-based antivirus infrastructure and I have some problems with WCCP resilience.

The WEB Caches are a couple of antivirus server. AV is TrendMicro IWSS 2.5 for windows and it is WCCP Compliant.

I am experiencing huge problemms with WCCP resilience. In particular if I disconnect the link of the router towards the AV servers (or I switch off it), when I reconnect it the two web caches (I.E. the AV Servers) are acquired but they get no buckets (the show ip wccp 80 detail shows a 0% hash allotment for both).

Please notice it happens both for standalone routers and HSRP.

Router is Cisco 2801. I tried different IOS releases, from 12.3(8T) to 12.4(5) with no results.

Pleae notice that if I unplug the interface not directly connected to the router, everything works fine.

Is this behaviour normal? Did someone face such a similar configuration.

Many Thanks

[Zach Seils]

The web caches (in this case the AV servers) are responsible for telling the router what the bucket assignments are. On the router you can do 'debug ip wccp packet' and look for Redirect_Assignment packets.

~Zach

[maxxo]

Hi Zach,

As a matter of fact I discovered that the redirect assign packet from the master cache (the one with the lowest ID) arrives too soon, that is before tha cache is aquired by the routers. The consequence is that the router discards the assignment with the following message:

WCCP-EVNT:D80: Redirect_Assignment packet from 10.11.3.18 fails source check

I have already contacted the vendor asking him to develop a feedback mechanism so that the cache keeps on sending the message until it realizes, from inspection of I_SEE_YOU messages, that the routers have received the assignment.

Meanwhile is there a workaround to avoid this situation?

Many thanks!

[Zach Seils]

Other than restarting the WCCP service, I do not know of a work around.

~Zach