In my case, simply adding Skylake-Client a supported CPU family did the trick: https://github.com/lago-project/lago/pull/448 i wonder if Westmere is a good fallback -- it works for you on Broadwell, right? On Wed, Feb 8, 2017 at 1:58 PM, Nadav Goldin <ngoldin@redhat.com> wrote:
I would first try testing it without OST, because in OST it will pick the CPU via the cluster family(which is controlled in virt.py). You can try specifying the 'cpu_model' in the init file, skipping the 'cpu family' logic, something like:
cat LagoInitFile domains: vm-el73: memory: 2048 service_provider: systemd cpu_model: Broadwell nics: - net: lago disks: - template_name: el7.3-base type: template name: root dev: vda format: qcow2 nets: lago: type: nat dhcp: start: 100 end: 254 management: true dns_domain_name: lago.local
lago init && lago start
Then install lago again in the VM, copy the same init file, and check if for different combinations of cpu_model it works for you - would give us a hint how to solve this. The 'cpu_model' basically translates to this xml definition in libvirt: <cpu mode='custom' match='exact'> <model fallback='allow'>Broadwell</model> <topology sockets='2' cores='1' threads='1'/> <feature policy='optional' name='vmx'/> <feature policy='optional' name='svm'/> </cpu>
I tried manually editing it also to host-passthrough, but still failed on the same error. The thing is that the 'kvm_put_msrs: Assertion `ret == n' failed.' error doesn't give any indication where it failed(or if the cpu is missing a flag), maybe there is a way to debug this at qemu/kvm level? I'm not sure.
It is a Skylake-H, and I can see it is not mentioned in lago/virt.py.
I guess I'll step through the code (as well as other places discovered by 'git grep cpu') and see if I could solve this by adding the Skylake family to _CPU_FAMILIES.
Do you have other pointers?
Thanks, Ondra
On Tue, Feb 7, 2017 at 10:40 PM, Nadav Goldin <ngoldin@redhat.com> wrote:
What is the host CPU you are using? I came across the same error few days ago, but without running OST, I tried running with Lago: fc24 host -> el7 vm -> el7 vm.
I have a slight suspect that it is related to the CPU model we configure in libvirt, I tried a mixture of few combinations(host-pass-through, pinning down the CPU model), but it always failed on the same error: kvm_put_msrs: Assertion `ret == n' failed.
My CPU is Broadwell btw.
Milan, any ideas? you think it might be related?
Nadav.
On Tue, Feb 7, 2017 at 11:14 PM, Ondrej Svoboda <osvoboda@redhat.com> wrote:
Yes, I stated that in my message.
root@osvoboda-t460p /home/src/ovirt-system-tests (git)-[master] # cat /sys/module/kvm_intel/parameters/nested :( Y
On Tue, Feb 7, 2017 at 1:39 PM, Eyal Edri <eedri@redhat.com> wrote:
Did you follow the instructions on [1] ?
Specifically, verifying ' cat /sys/module/kvm_intel/
' gives you 'Y'.
[1]
http://ovirt-system-tests.readthedocs.io/en/latest/docs/ general/installation.html
On Tue, Feb 7, 2017 at 2:29 PM, Ondrej Svoboda <osvoboda@redhat.com> wrote:
Hi everyone,
Even though I have nested virtualization enabled in my Arch Linux system which I use to run OST, vm_run is the first test to fail in 004_basic_sanity (followed by snapshots_merge and suspend_resume_vm).
Can you point me to what I might be missing? I believe I get the
same
failure even on Fedora.
This is what host0's CPU capabilities look like (vmx is there): [root@lago-basic-suite-master-host0 ~]# cat /proc/cpuinfo processor : 0 vendor_id : GenuineIntel cpu family : 6 model : 44 model name : Westmere E56xx/L56xx/X56xx (Nehalem-C) stepping : 1 microcode : 0x1 cpu MHz : 2711.988 cache size : 16384 KB physical id : 0 siblings : 1 core id : 0 cpu cores : 1 apicid : 0 initial apicid : 0 fpu : yes fpu_exception : yes cpuid level : 11 wp : yes flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 syscall nx lm constant_tsc rep_good nopl xtopology pni pclmulqdq vmx ssse3 cx16 sse4_1 sse4_2 x2apic popcnt aes hypervisor lahf_lm arat tpr_shadow vnmi flexpriority ept vpid bogomips : 5423.97 clflush size : 64 cache_alignment : 64 address sizes : 40 bits physical, 48 bits virtual power management:
journalctl -b on host0 shows that libvirt complains about NUMA configuration:
Feb 07 06:33:23 lago-basic-suite-master-host0 libvirtd[12888]:
version: 2.0.0, package: 10.el7_3.4 (CentOS BuildSystem <http://bugs.centos.org>, 2017-01-17-23:37:48, c1bm.rdu2.centos.org ) Feb 07 06:33:23 lago-basic-suite-master-host0 kernel: ovirtmgmt:
2(vnet0) entered disabled state Feb 07 06:33:23 lago-basic-suite-master-host0 kernel: device vnet0 left promiscuous mode Feb 07 06:33:23 lago-basic-suite-master-host0 kernel: ovirtmgmt:
On Wed, Feb 8, 2017 at 1:18 PM, Ondrej Svoboda <osvoboda@redhat.com> wrote: parameters/nested libvirt port port
2(vnet0) entered disabled state Feb 07 06:33:23 lago-basic-suite-master-host0 libvirtd[12888]: hostname: lago-basic-suite-master-host0.lago.local Feb 07 06:33:23 lago-basic-suite-master-host0 libvirtd[12888]: Unable to read from monitor: Connection reset by peer Feb 07 06:33:23 lago-basic-suite-master-host0 libvirtd[12888]: internal error: qemu unexpectedly closed the monitor: 2017-02-07T11:33:23.058571Z qemu-kvm: warning: CPU(s) not present in any NUMA nodes: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
2017-02-07T11:33:23.058826Z qemu-kvm: warning: All CPU(s) up to maxcpus should be described in NUMA config
qemu-kvm: /builddir/build/BUILD/qemu-2.6.0/target-i386/kvm.c:1736: kvm_put_msrs: Assertion `ret == n' failed. Feb 07 06:33:23 lago-basic-suite-master-host0 NetworkManager[657]: <info> [1486467203.1025] device (vnet0): state change: disconnected -> unmanaged (reason 'unmanaged') [30 10 3] Feb 07 06:33:23 lago-basic-suite-master-host0 kvm[22059]: 0 guests now active Feb 07 06:33:23 lago-basic-suite-master-host0 systemd-machined[22044]: Machine qemu-1-vm0 terminated.
Thanks, Ondra
_______________________________________________ Devel mailing list Devel@ovirt.org http://lists.ovirt.org/mailman/listinfo/devel
-- Eyal Edri Associate Manager RHV DevOps EMEA ENG Virtualization R&D Red Hat Israel
phone: +972-9-7692018 irc: eedri (on #tlv #rhev-dev #rhev-integ)
_______________________________________________ Devel mailing list Devel@ovirt.org http://lists.ovirt.org/mailman/listinfo/devel