Installed the qemu-kvm-ev but the behavior is the same. Running perf
record -a -g on the baremetal shows that most of the CPU time is in
_raw_spin_lock
Children Self Command Shared Object
Symbol
- 93.62% 93.62% qemu-kvm [kernel.kallsyms] [k]
_raw_spin_lock
- _raw_spin_lock
+ 45.30% kvm_mmu_sync_roots
+ 28.49% kvm_mmu_load
+ 25.00% mmu_free_roots
+ 1.12% tdp_page_fault
On Wed, Aug 17, 2016 at 7:35 PM, Laurentiu Soica <laurentiu@xxxxxxxx> wrote:
> Hello,
>
> I have an openstack setup with KVM as hipervisor in virtual environment.
>
> So:
> - a baremetal with 2 physical CPUs: 128 GB RAM
> - the compute node, that is a VM under KVM on the baremetal, with 100
> GB RAM and 36 vCPUs
> - 15 VMs running inside the compute node, under KVM aswell.
>
> The baremetal and compute are both running CentOS 7 and on the
> baremetal I've enabled nested KVM feature.
> Kernel is 3.10.0-327.22.2.el7.x86_64
> qemu-kvm is 1.5.3-105.el7_2.7
>
> The issue is that after a few days of running, the compute node's
> qemu-kvm process (on the baremetal) goes into full CPU usage (3600%)
> and the VMs under it are no longer accessible.
> The swap is disabled on the compute and there is no swap activity on
> baremetal as well (aroung 100 MB used).
>
> When the compute goes into full CPU usage the baremetal has around
> 60GB of RAM still free and the compute around 70 GB of RAM free.
>
> If something rings a bell or do you have some troubleshooting tips,
> please let me know.
>
> Laurentiu
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
