There have been a number of failures in various causal consistency suites where we're seeing commands time out during periods of 100% disk utilization on the RHEL 6.2 variant. We're currently running 3-4 jobs in each of the suites, and we should lower the number of jobs at least for this variant.