On 10 Feb 2017, at 10:26, Michal Skrivanek <mskrivan@redhat.com> wrote:
On 9 Feb 2017, at 16:16, Ondrej Svoboda <osvoboda@redhat.com <mailto:osvoboda@redhat.com>> wrote:
Do you mean https://github.com/lago-project/lago/pull/398 <https://github.com/lago-project/lago/pull/398> which has been merged for over a month?
The second sentence in the PR (below) is contradicted by newer, non-recognized CPUs, such as Skylake.
How/why? Westmere should have been selected in that case
And if it was and didn’t work for you then it is a nested virtualization compatibility bug you should report to QEMU/KVM folks.
"This patch fixes the problems by selecting a minimum reasonable CPU model for the given hardware platform. Westmere is selected unless older or non-Intel hardware is used."
On Thu, Feb 9, 2017 at 4:07 PM, Michal Skrivanek <mskrivan@redhat.com <mailto:mskrivan@redhat.com>> wrote: What happened to Milan' PR from a while ago addressing this exact situation?
On 08 Feb 2017, at 16:04, Ondrej Svoboda <osvoboda@redhat.com <mailto:osvoboda@redhat.com>> wrote:
In my case, simply adding Skylake-Client a supported CPU family did the trick: https://github.com/lago-project/lago/pull/448 <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 <mailto: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.
On Wed, Feb 8, 2017 at 1:18 PM, Ondrej Svoboda <osvoboda@redhat.com <mailto:osvoboda@redhat.com>> wrote:
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 <mailto: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 <mailto: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 <mailto:eedri@redhat.com>> wrote: > > Did you follow the instructions on [1] ? > > Specifically, verifying ' cat /sys/module/kvm_intel/parameters/nested > ' > gives you 'Y'. > > [1] > > http://ovirt-system-tests.readthedocs.io/en/latest/docs/general/installation... <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 <mailto: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]: libvirt >> version: 2.0.0, package: 10.el7_3.4 (CentOS BuildSystem >> <http://bugs.centos.org <http://bugs.centos.org/>>, 2017-01-17-23 <tel:2017-01-17-23>:37:48, c1bm.rdu2.centos.org <http://c1bm.rdu2.centos.org/>) >> Feb 07 06:33:23 lago-basic-suite-master-host0 kernel: ovirtmgmt: port >> 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: 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 <mailto:Devel@ovirt.org> >> http://lists.ovirt.org/mailman/listinfo/devel <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 <tel:%2B972-9-7692018> > irc: eedri (on #tlv #rhev-dev #rhev-integ)
_______________________________________________ Devel mailing list Devel@ovirt.org <mailto:Devel@ovirt.org> http://lists.ovirt.org/mailman/listinfo/devel <http://lists.ovirt.org/mailman/listinfo/devel>
_______________________________________________ Devel mailing list Devel@ovirt.org <mailto:Devel@ovirt.org> http://lists.ovirt.org/mailman/listinfo/devel <http://lists.ovirt.org/mailman/listinfo/devel>
Devel mailing list Devel@ovirt.org <mailto:Devel@ovirt.org> http://lists.ovirt.org/mailman/listinfo/devel <http://lists.ovirt.org/mailman/listinfo/devel>