PBR and next-hop outage

[Rick Morris]

In the PBR configuration is there something specific needed to assure traffic to continue to flow if the next-hop is down?

For instance, if I am sending all www traffic through a cache and the server dies, how do I set this up so traffic can flow?  The cache vendor is stating that it will just bypass the PBR, I am not a PBR expert but do not believe this to be correct.  If I send traffic there then it will always attempt to route that direction unless specified.  Am I correct?  How is this handled?

[Giuseppe Larosa · Accepted solution]

Hello Rick,

you are correct.

The PBR action will be changed and reverted to standard destination routing if the device detects the next-hop failure.

To improve next-hop failure detection there are some ways

set ip next-hop verify-availability   

in the PBR route-map if supported on the device

next-hop should be on a directly connected SVI if I correctly remember.

there is also a newer implementation that uses IP SLA  and tracking with PBR

see

http://www.cisco.com/en/US/docs/ios/12_3t/12_3t4/feature/guide/gtpbrtrk.html

To be noted for web caches the best is to use WCCP that provides cache detection via WCCP hellos, but it needs to be supported also on the web cache.

Hope to help

Giuseppe