Hi all and happy new year! I'm running oVirt 3.6 on EL7. In order to test nested-kvm I installed vdsm-hook-nestedvt on the hypervisor and put "fake_kvm_support = true" in /etc/vdsm/vdsm.conf.

However, afterwards the hypervisor host cannot connect to the ovirt Manager, on ovirt Manager the engine.log will tell: 2016-01-01 15:21:42,105 WARN [org.ovirt.engine.core.dal.dbbroker.auditloghandling.AuditLogDirector] (DefaultQuartzScheduler_Worker-76) [71d1638c] Correlation ID: 29daf92c, Call Stack: null, Custom Event ID: -1, Message: Host bigvirt moved to Non-Operational state as host does not meet the cluster's minimum CPU level. Missing CPU features : model_Opteron_G3 After removing the option all works well again, but I'ḿ lacking the nested kvm option :( Is nested kvm only supported on a limited cpu's?? Winny Information from /proc/cpuinfo: processor : 0 vendor_id : AuthenticAMD cpu family : 16 model : 10 model name : AMD Phenom(tm) II X6 1055T Processor stepping : 0 microcode : 0x10000bf cpu MHz : 2800.000 cache size : 512 KB physical id : 0 siblings : 6 core id : 0 cpu cores : 6 apicid : 0 initial apicid : 0 fpu : yes fpu_exception : yes cpuid level : 6 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 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm 3dnowext 3dnow constant_tsc rep_good nopl nonstop_tsc extd_apicid aperfmperf pni monitor cx16 popcnt lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs skinit wdt cpb hw_pstate npt lbrv svm_lock nrip_save pausefilter bogomips : 5625.54 TLB size : 1024 4K pages clflush size : 64 cache_alignment : 64 address sizes : 48 bits physical, 48 bits virtual power management: ts ttp tm stc 100mhzsteps hwpstate cpb _______________________________________________ Users mailing list Users(a)ovirt.org http://lists.ovirt.org/mailman/listinfo/users

On 04/01/16 09:52 +0100, Simone Tiraboschi wrote: > On Fri, Jan 1, 2016 at 3:32 PM, Winfried de Heiden <wdh(a)dds.nl&gt; wrote: > >> Hi all and happy new year! >> >> I'm running oVirt 3.6 on EL7. >> >> In order to test nested-kvm I installed vdsm-hook-nestedvt on the >> hypervisor and put "fake_kvm_support = true" in /etc/vdsm/vdsm.conf. > > > > > fake_kvm_support is just for developing purposes and it's not needed > as it > prevent guests to run. Not really true - fake_kvm_support enables QEMU emulation instead of hardware virtualization (e.g. VT-x) instructions. It is also quite broken in it's current state and fix is pending in cpuinfo branch at gerrit. I don't really understand the reason why fake_kvm would be used in this case as AMD-v is enabled according to svm flag present - following instructions should be correct for getting the flag passed to the VMs. > Please check to understand if nested support is enabled on your host: > /sys/module/kvm_intel/parameters/nested > > if not > echo "options kvm-intel nested=1" > /etc/modprobe.d/kvm-intel.conf > and reload the module > Than install vdsm-hook-nestedvt on your hosts. > If you want also to use the network on your L2 guest please install > and > configure also macspoof vdsm hook on your physical hosts otherwise all > the > traffic to/from your L2 guests will be dropped. > https://github.com/oVirt/vdsm/tree/master/vdsm_hooks/macspoof > > > > >> However, afterwards the hypervisor host cannot connect to the ovirt >> Manager, on ovirt Manager the engine.log will tell: >> >> 2016-01-01 15:21:42,105 WARN >> [org.ovirt.engine.core.dal.dbbroker.auditloghandling.AuditLogDirector] >> (DefaultQuartzScheduler_Worker-76) [71d1638c] Correlation ID: >> 29daf92c, >> Call Stack: null, Custom Event ID: -1, Message: Host bigvirt moved to >> Non-Operational state as host does not meet the cluster's minimum CPU >> level. Missing CPU features : model_Opteron_G3 >> >> After removing the option all works well again, but I'ḿ lacking the >> nested >> kvm option :( Is nested kvm only supported on a limited cpu's?? >> >> Winny >> >> Information from /proc/cpuinfo: >> >> processor : 0 >> vendor_id : AuthenticAMD >> cpu family : 16 >> model : 10 >> model name : AMD Phenom(tm) II X6 1055T Processor >> stepping : 0 >> microcode : 0x10000bf >> cpu MHz : 2800.000 >> cache size : 512 KB >> physical id : 0 >> siblings : 6 >> core id : 0 >> cpu cores : 6 >> apicid : 0 >> initial apicid : 0 >> fpu : yes >> fpu_exception : yes >> cpuid level : 6 >> 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 ht syscall nx mmxext >> fxsr_opt >> pdpe1gb rdtscp lm 3dnowext 3dnow constant_tsc rep_good nopl >> nonstop_tsc >> extd_apicid aperfmperf pni monitor cx16 popcnt lahf_lm cmp_legacy svm >> extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs >> skinit wdt >> cpb hw_pstate npt lbrv svm_lock nrip_save pausefilter >> bogomips : 5625.54 >> TLB size : 1024 4K pages >> clflush size : 64 >> cache_alignment : 64 >> address sizes : 48 bits physical, 48 bits virtual >> power management: ts ttp tm stc 100mhzsteps hwpstate cpb >> >> _______________________________________________ >> Users mailing list >> Users(a)ovirt.org >> http://lists.ovirt.org/mailman/listinfo/users >> > _______________________________________________ > Users mailing list > Users(a)ovirt.org > http://lists.ovirt.org/mailman/listinfo/users

From: "Martin Polednik" <mpolednik(a)redhat.com&gt; To: wdh(a)dds.nl Cc: "users" <users(a)ovirt.org&gt;, "Martin Polednik" <mpolednik(a)redhat.com&gt; Sent: Monday, January 4, 2016 2:52:03 PM Subject: Re: [ovirt-users] Nested KVM - vdsm-hook-nestedvt On 04/01/16 14:38 +0100, wdh(a)dds.nl wrote: >Hi all, > >OK, it seems I am confused. Nested KVM is NOT the same as fake KVM.... > >If so, I installed vdsm-hook-nestedvt on the hypervisor (only). Trying >to run ovirt-Node on a VM it complains there 's no virtualization, >despite the kvm_amd module is loaded with nested=1 > >How to enable nested-kvm for oVirt 3.6? First, make sure that you're enabling the nested feature for amd (just noticed that the example used intel). Please show us the output of # cat /sys/module/kvm_intel/parameters/nested

on the host (physical machine), lscpu inside the VM and VM XML - use # virsh -r list # virsh -r dumpxml {machine_id found in the list} Thanks, mpolednik >Winny > > >Martin Polednik schreef op 04-01-2016 13:29: >>On 04/01/16 09:52 +0100, Simone Tiraboschi wrote: >>>On Fri, Jan 1, 2016 at 3:32 PM, Winfried de Heiden <wdh(a)dds.nl&gt; wrote: >>> >>>>Hi all and happy new year! >>>> >>>>I'm running oVirt 3.6 on EL7. >>>> >>>>In order to test nested-kvm I installed vdsm-hook-nestedvt on the >>>>hypervisor and put "fake_kvm_support = true" in /etc/vdsm/vdsm.conf. >>> >>> >>> >>> >>>fake_kvm_support is just for developing purposes and it's not >>>needed as it >>>prevent guests to run. >> >>Not really true - fake_kvm_support enables QEMU emulation instead of >>hardware virtualization (e.g. VT-x) instructions. It is also quite >>broken in it's current state and fix is pending in cpuinfo branch at >>gerrit. >> >>I don't really understand the reason why fake_kvm would be used in >>this case as AMD-v is enabled according to svm flag present - >>following instructions should be correct for getting the flag passed >>to the VMs. >> >>>Please check to understand if nested support is enabled on your host: >>>/sys/module/kvm_intel/parameters/nested >>> >>>if not >>> echo "options kvm-intel nested=1" > /etc/modprobe.d/kvm-intel.conf >>>and reload the module >> >>>Than install vdsm-hook-nestedvt on your hosts. >>>If you want also to use the network on your L2 guest please >>>install and >>>configure also macspoof vdsm hook on your physical hosts otherwise >>>all the >>>traffic to/from your L2 guests will be dropped. >>>https://github.com/oVirt/vdsm/tree/master/vdsm_hooks/macspoof >>> >>> >>> >>> >>>>However, afterwards the hypervisor host cannot connect to the ovirt >>>>Manager, on ovirt Manager the engine.log will tell: >>>> >>>>2016-01-01 15:21:42,105 WARN >>>>[org.ovirt.engine.core.dal.dbbroker.auditloghandling.AuditLogDirector] >>>>(DefaultQuartzScheduler_Worker-76) [71d1638c] Correlation ID: >>>>29daf92c, >>>>Call Stack: null, Custom Event ID: -1, Message: Host bigvirt moved to >>>>Non-Operational state as host does not meet the cluster's minimum CPU >>>>level. Missing CPU features : model_Opteron_G3 >>>> >>>>After removing the option all works well again, but I'ḿ lacking >>>>the nested >>>>kvm option :( Is nested kvm only supported on a limited cpu's?? >>>> >>>>Winny >>>> >>>>Information from /proc/cpuinfo: >>>> >>>>processor : 0 >>>>vendor_id : AuthenticAMD >>>>cpu family : 16 >>>>model : 10 >>>>model name : AMD Phenom(tm) II X6 1055T Processor >>>>stepping : 0 >>>>microcode : 0x10000bf >>>>cpu MHz : 2800.000 >>>>cache size : 512 KB >>>>physical id : 0 >>>>siblings : 6 >>>>core id : 0 >>>>cpu cores : 6 >>>>apicid : 0 >>>>initial apicid : 0 >>>>fpu : yes >>>>fpu_exception : yes >>>>cpuid level : 6 >>>>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 ht syscall nx mmxext >>>>fxsr_opt >>>>pdpe1gb rdtscp lm 3dnowext 3dnow constant_tsc rep_good nopl >>>>nonstop_tsc >>>>extd_apicid aperfmperf pni monitor cx16 popcnt lahf_lm cmp_legacy svm >>>>extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw ibs >>>>skinit wdt >>>>cpb hw_pstate npt lbrv svm_lock nrip_save pausefilter >>>>bogomips : 5625.54 >>>>TLB size : 1024 4K pages >>>>clflush size : 64 >>>>cache_alignment : 64 >>>>address sizes : 48 bits physical, 48 bits virtual >>>>power management: ts ttp tm stc 100mhzsteps hwpstate cpb >>>> >>>>_______________________________________________ >>>>Users mailing list &gt;&gt;&gt;&gt;Users(a)ovirt.org >>>>http://lists.ovirt.org/mailman/listinfo/users >>>> >> >>>_______________________________________________ >>>Users mailing list &gt;&gt;&gt;Users(a)ovirt.org >>>http://lists.ovirt.org/mailman/listinfo/users _______________________________________________ Users mailing list Users(a)ovirt.org http://lists.ovirt.org/mailman/listinfo/users

Sounds like: https://bugzilla.redhat.com/show_bug.cgi?id=1276651

Can you post exact flow and logs? Thanks.

On Sun, Jan 03, 2016 at 03:18:40PM +0100, Frank Wall wrote: > [ INFO ] Checking for product updates... > [ ERROR ] Yum: [u'ovirt-engine-3.6.1.3-1.el6.noarch requires slf4j >= 1.7.0', u'vdsm-jsonrpc-java-1.1.5-1.el6.noarch requires slf4j >= 1.6.1'] > [ INFO ] Yum: Performing yum transaction rollback > [ ERROR ] Failed to execute stage 'Environment customization': [u'ovirt-engine-3.6.1.3-1.el6.noarch requires slf4j >= 1.7.0', u'vdsm-jsonrpc-java-1.1.5-1.el6.noarch requires slf4j >= 1.6.1'] Nobody's using EL6 anymore? :)

I'd switch my Hosted-Engine VM to EL7 if it wouldn't be such a pain. The only "Upgrade Guide" I could find for Engine VM is this: http://www.ovirt.org/OVirt_Hosted_Engine_Backup_and_Restore No fun. :(

From: "Frank Wall" <fw(a)moov.de&gt; To: "Yedidyah Bar David" <didi(a)redhat.com&gt; Cc: "users" <users(a)ovirt.org&gt; Sent: Tuesday, January 5, 2016 5:47:51 PM Subject: Re: [ovirt-users] Unable to upgrade ovirt-engine 3.5.5 to 3.6.1 on EL6 On Tue, Jan 05, 2016 at 05:37:08PM +0200, Yedidyah Bar David wrote: > Perhaps to aid debugging you can try this, but _do_not_ press Enter: > > yum --disableplugin=versionlock update ovirt-engine > > Kill it (or reply 'no'). > > What does it output? Basically the same dependency error. I've attached the full output to this e-mail. To answer your other questions: > 16:43 <didi> fraenki: what happens when you try yum update slf4j? It is not currently installed (on oVirt 3.5.5) and only the inappropiate version is available: # yum --showduplicates list slf4j [...] Available Packages slf4j.noarch 1.5.8-8.el6 centos-base > 16:44 <didi> fraenki: also for vdsm-jsonrpc-java # yum update vdsm-jsonrpc-java [...] --> Running transaction check ---> Package vdsm-jsonrpc-java.noarch 0:1.0.15-1.el6 will be updated --> Processing Dependency: vdsm-jsonrpc-java < 1.1.0 for package: ovirt-engine-backend-3.5.5-1.el6.noarch ---> Package vdsm-jsonrpc-java.noarch 0:1.1.5-1.el6 will be an update --> Processing Dependency: slf4j >= 1.6.1 for package: vdsm-jsonrpc-java-1.1.5-1.el6.noarch [...] Error: Package: vdsm-jsonrpc-java-1.1.5-1.el6.noarch (ovirt-3.6) Requires: slf4j >= 1.6.1 Available: slf4j-1.5.8-8.el6.noarch (centos-base) slf4j = 1.5.8-8.el6 [...] Very similar (of not the same) output to the `yum update ovirt-engine` command. Where should I get "slf4j >= 1.6.1" from? It does not seem to be available from the "centos-base" repository (there's only slf4j = 1.5.8-8.el6 available): http://mirror.centos.org/centos/6/os/x86_64/Packages/ > 16:46 <didi> fraenki: if it still fails for you after the usual things, > e.g. > yum clean all, check repos etc., I'll try to reproduce Yeah, tried this. Already wiped my repos and reinstalled ovirt-release36.rpm. FWIW, I've noticed that a more recent slf4j is available from the SL6 repo at: http://ftp.scientificlinux.org/linux/scientific/6x/external_products/mave... But I fear that this will break things in the long run (and it would replace several other packages as well, very likely to break ovirt-engine completely). > > I'd switch my Hosted-Engine VM to EL7 if it wouldn't be such a pain. The > > only > > "Upgrade Guide" I could find for Engine VM is this: > > http://www.ovirt.org/OVirt_Hosted_Engine_Backup_and_Restore > > No fun. :( > > This one does not deal with upgrading the engine OS. For that Sandro > started [1] but gave up after having some issues. Yeah, I wasn't really expecting that an in-place upgrade EL6->EL7 would work in this case. That's why I was thinking about the Backup/Restore method... Regards - Frank _______________________________________________ Users mailing list Users(a)ovirt.org http://lists.ovirt.org/mailman/listinfo/users

On Tue, Jan 05, 2016 at 05:37:08PM +0200, Yedidyah Bar David wrote: > Perhaps to aid debugging you can try this, but _do_not_ press Enter: > > yum --disableplugin=versionlock update ovirt-engine > > Kill it (or reply 'no'). > > What does it output? Basically the same dependency error. I've attached the full output to this e-mail. To answer your other questions: > 16:43 <didi> fraenki: what happens when you try yum update slf4j? It is not currently installed (on oVirt 3.5.5) and only the inappropiate version is available:

# yum --showduplicates list slf4j [...] Available Packages slf4j.noarch 1.5.8-8.el6 centos-base > 16:44 <didi> fraenki: also for vdsm-jsonrpc-java # yum update vdsm-jsonrpc-java [...] --> Running transaction check ---> Package vdsm-jsonrpc-java.noarch 0:1.0.15-1.el6 will be updated --> Processing Dependency: vdsm-jsonrpc-java < 1.1.0 for package: ovirt-engine-backend-3.5.5-1.el6.noarch ---> Package vdsm-jsonrpc-java.noarch 0:1.1.5-1.el6 will be an update --> Processing Dependency: slf4j >= 1.6.1 for package: vdsm-jsonrpc-java-1.1.5-1.el6.noarch [...] Error: Package: vdsm-jsonrpc-java-1.1.5-1.el6.noarch (ovirt-3.6) Requires: slf4j >= 1.6.1 Available: slf4j-1.5.8-8.el6.noarch (centos-base) slf4j = 1.5.8-8.el6 [...] Very similar (of not the same) output to the `yum update ovirt-engine` command. Where should I get "slf4j >= 1.6.1" from?

It does not seem to be available from the "centos-base" repository (there's only slf4j = 1.5.8-8.el6 available): http://mirror.centos.org/centos/6/os/x86_64/Packages/ > 16:46 <didi> fraenki: if it still fails for you after the usual things, e.g. > yum clean all, check repos etc., I'll try to reproduce Yeah, tried this. Already wiped my repos and reinstalled ovirt-release36.rpm.

FWIW, I've noticed that a more recent slf4j is available from the SL6 repo at: http://ftp.scientificlinux.org/linux/scientific/6x/external_products/mave... But I fear that this will break things in the long run (and it would replace several other packages as well, very likely to break ovirt-engine completely). > > I'd switch my Hosted-Engine VM to EL7 if it wouldn't be such a pain. The only > > "Upgrade Guide" I could find for Engine VM is this: > > http://www.ovirt.org/OVirt_Hosted_Engine_Backup_and_Restore > > No fun. :( > > This one does not deal with upgrading the engine OS. For that Sandro > started [1] but gave up after having some issues. Yeah, I wasn't really expecting that an in-place upgrade EL6->EL7 would work in this case. That's why I was thinking about the Backup/Restore method...