On Fri, Oct 16, 2015 at 2:23 PM, Marius Millea <[email protected]> wrote:
> Here's the relevant lines from the scheduler log, vmx is there, and its > present in /proc/cpuinfo as well. > > 2015-10-16 12:12:34.6279 [PID=66 ] [send] CPU features: fpu vme de > pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts > acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx rdtscp lm constant_tsc > arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc aperfmperf > eagerfpu pni pclmulqdq dtes64 monitor ds_cpl *vmx* smx est tm2 ssse3 cx16 > xtpr pdcm pcid sse4_1 sse4_2 x2apic popcnt tsc_deadline_timer aes xsave avx > f16c rdrand lahf_lm ida arat epb xsaveopt pln pts dtherm tpr_shadow vnmi > flexpriority ept vpid fsgsbase smep erms > 2015-10-16 12:12:34.6596 [PID=66 ] [send] Sending app_version > camb_boinc2docker 4 5 vbox64_mt; projected 14.64 GFLOPS > > Reading eg here > <http://www.howtogeek.com/howto/linux/linux-tip-how-to-tell-if-your-processor-supports-vt/> > or here > <http://askubuntu.com/questions/103965/how-to-determine-if-cpu-vt-extensions-enabled-in-bios> > it > seems vmx appearing there just means your CPU supports it, not that it is > actually enabled. Clearly in my case its disabled yet still shows up. > > So I suppose the questions are now, > > 1) is there a way BOINC can check if vmx is actually enabled? > 2) whats going on with vboxwrapper > > Yea, can confirm "VBoxManage startvm" returns successfully on my system even though the VM immediately dies with the VERR_VMX_MSR_VMXON_DISABLED error. That's why vboxwrapper never catches it. Not sure if this is a VBox bug and/or if its limited to my system. > Marius > > > > On Fri, Oct 16, 2015 at 2:01 PM, Christian Beer <[email protected] > > wrote: > >> Did you restart the server processes just to make sure? The scheduler >> should use the new plan_class.xml config instantly and not send you any >> work. This seems to be a separate problem. >> >> So in theory the server shouldn't send you any task but it is. And the >> vboxwrapper should recognize the error code but it isn't. Very strange. >> >> Next thing to check would be the scheduler_request and reply from the >> client and the scheduler log on the server the interesting part there is >> the cpu capabilities. This is where the BIOS setting is reported. There >> should be no vmx or svm capability reported by your computer (as you >> have HW accell turned off). >> >> Regards >> Christian >> >> On 10/16/2015 01:47 PM, Marius Millea wrote: >> > Thanks Christian, I added that option but same result. >> > >> > Its clear to me now that vboxwrapper is missing the fact that the VM >> > doesn't start correctly. If you look at the attached stderr.txt, if >> > says "Successfully started VM." Hence it never makes it to the part in >> > the code that checks vbox.log for "VERR_VMX_MSR_VMXON_DISABLED" and >> > hence never prints "ERR_CPU_VM_EXTENSIONS_DISABLED". >> > >> > I'm looking into why, any ideas? Otherwise yea it looks like the code >> > to do exactly what you guys describe is there. >> > >> > Marius >> > >> > >> > On Fri, Oct 16, 2015 at 11:42 AM, Christian Beer >> > <[email protected] <mailto:[email protected]>> wrote: >> > >> > On 10/16/2015 11:16 AM, Marius Millea wrote: >> > > However I'm using VBox >> > > 4.3.30 and vboxwrapper 26175 so I don't think that's it. I'm >> > attaching my >> > > vbox.log, there's some errors in there about VT-x but searching >> > > for ERR_CPU_VM_EXTENSIONS_DISABLED yields nothing. >> > The vbox.log contains VERR_VMX_MSR_VMXON_DISABLED at the end. Which >> is >> > the VirtualBox errorcode to say you have a 64bit Guest but Hardware >> > Virtualization is not enabled (it says so in the begining of the log >> > too). This gets picked up by vboxwrapper and >> > ERR_CPU_VM_EXTENSIONS_DISABLED gets output to stderr.txt in the slot >> > directory which gets reported back to the server as part of the >> > sched_request file. The stderr.txt shoud be visible as part of the >> > failed task once it is reported. >> > >> > You should change the plan_class vbox64 to include the >> > <vm_accel_required/> tag. This is the plan_class I use at RNA World: >> > > <plan_class> >> > > <name>vbox64</name> >> > > <virtualbox/> >> > > <is64bit/> >> > > <min_core_client_version>70224</min_core_client_version> >> > > <min_vbox_version>40010</min_vbox_version> >> > > <vm_accel_required/> >> > > </plan_class> >> > This will not send a 64bit VM to a host that has no Hardware >> > Virtualization enabled. This should really be the default as >> > VirtualBox >> > doesn't allow a 64bit guest without the HW Virtualization enabled. >> > >> > Regards >> > Christian >> > _______________________________________________ >> > boinc_dev mailing list >> > [email protected] <mailto:[email protected]> >> > http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev >> > To unsubscribe, visit the above URL and >> > (near bottom of page) enter your email address. >> > >> > >> >> _______________________________________________ >> boinc_dev mailing list >> [email protected] >> http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev >> To unsubscribe, visit the above URL and >> (near bottom of page) enter your email address. >> > > _______________________________________________ boinc_dev mailing list [email protected] http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev To unsubscribe, visit the above URL and (near bottom of page) enter your email address.
