[impala-user] Impala on Llama with YARN

I would like to know if anyone has any advice on how to configure the Dynamic Resource Pools for Impala integrated with YARN using Llama. Every once in a while, we get a timeout exception waiting for resource expansion by Llama. It only happens when the load on the cluster goes up. Right now, Impala and other internal data processing apps share the same higher weighted resource pool. If anyone out there has went thru configuring and tuning this setup, I would truly appreciate if you can share what you have done.