EL6 CPU Limit Under Xen

Latest response

I'm working on a project where we're migrating a number of tenants from physically- and VM-ware hosted systems into AWS. In prep for that task, I was reading through the EC2 documentation so that I'd be able to help our tenants appropriately size their target instances to match their current and expected workload. While reading through the documentation for C4 instances, I noticed a note stating that EL6 wasn't capable of supporting more than 32 CPUs. This surprised me, given that we have a several EL6 tenants that are using systems that exceed this number and the RedHat limits matrix only showed a 32 CPU limit for 32-bit EL6, but not 64bit. So, I wrote to the maintainers of the AWS C4 instances to get the scoop.

They replied back that the limit was specific to Xen-hosted EL6 deployments, that Red Hat was aware of the problem and that a solution was being worked out. I went back and checked the limits matrix to verify that I'd not mist an asterisk or footnote. There was none to be found and even a search of the page for "xen" failed to score a hit. Does anyone have any insight into the specific problem? Even more, does Red Hat have an anticipated fix time-frame (I need to be able to tell my tenant "you guys will need to either hold on re-deploying this component or figure out if you can do a scaled-out deployment of it in its stead")?

Responses

Private Bug 1164278 - Xen: Add support for >32 VCPUs is going into RHEL 6.7, and is being backported to RHEL 6.6.z under Private Bug 1179343.

QE testing has shown 64-bit kernels work fine with larger VCPU numbers, I'm not quite clear on the results but it seems 32-bit kernels still have a limit of 32 VCPUs. I've asked for clarification of these results/limits on the main bug.

Release target for the 6.6.z kernel is mid-March.

Close

Welcome! Check out the Getting Started with Red Hat page for quick tours and guides for common tasks.