> On Jun 29, 2016, at 3:00 PM, Geoff Nordli <geo...@gnaa.net> wrote: > > > > On 2016-06-28 06:18 PM, Dale Ghent wrote: >>> On Jun 28, 2016, at 8:25 PM, Geoff Nordli <geo...@gnaa.net> wrote: >>> >>> >>> >>> On 2016-06-28 12:57 PM, Geoff Nordli wrote: >>>> >>>> On 2016-06-28 12:23 PM, Dale Ghent wrote: >>>>> Is the vbox version you’re using on your new server different from that >>>>> which you are using on existing ones? >>>>> >>>>> Are you compiling vbox on OmniOS, or using the pre-compiled Solaris >>>>> binaries from Oracle? >>>>> >>>>> /dale >>>>> >>>>>> On Jun 28, 2016, at 3:11 PM, Geoff Nordli <geo...@gnaa.net> wrote: >>>>>> >>>>>> >>>>>> >>>>>> On 2016-06-28 11:59 AM, John D Groenveld wrote: >>>>>>> In message <5772c723.80...@gnaa.net>, Geoff Nordli writes: >>>>>>>> I am running OmniOS v11 r151018. >>>>>>>> >>>>>>>> I have quite a few OI/Omnios OS boxes running Virtualbox. I haven't see >>>>>>>> this before. On a new server I just installed I am getting a memory >>>>>>>> fault right away when starting a vm. >>>>>>> Perhaps set disable_smap=1? >>>>>>> <URL:http://openindiana.org/pipermail/openindiana-discuss/2016-June/019049.html> >>>>>>> >>>>>>> BTW have you done any benchmarking between VBox and KVM? >>>>>>> John >>>>>>> groenv...@acm.org >>>>>>> >>>>>> Hi John. >>>>>> >>>>>> Yes, I set the disable_smap in the system and rebooted, but it doesn't >>>>>> seem to help. That setting supposedly prevents the host from rebooting. >>>>>> >>>>>> On a Windows Server 2012 R2 I am getting the following networking >>>>>> performance with iperf (default settings). >>>>>> >>>>>> Client -> VM 903Mbps >>>>>> VM -> Client 572Mbps >>>>>> >>>>>> I am focused on small business systems and protecting their data, so >>>>>> these numbers are more than good enough. >>>>>> >>>>>> >>>> Hi Dale. >>>> >>>> I am using the pre-compiled binaries from the vbox download site. >>>> >>>> I have another server using the same OmniOS release. The one that works >>>> is running 5.0.20, but i have tried both versions of vbox on the new >>>> server and both are producing the memory fault. >>>> >>>> The only thing that seems to be different about these boxes are the CPUs. >>>> The one that works has older Xeons, the new one that doesn't work is >>>> running E5-2630V3 (haswell). >>>> >>>> thanks, >>>> >>>> Geoff >>>> >>>> >>> Just tried the newly released 5.0.24 and it still gives the same error. >> Well, using Solaris on illumos is going to be a more and more iffy >> proposition as time marches on; it's hard to say what might have changed on >> Oracle Solaris, if anything; that the vbox packages were compiled on. >> >> How about trying a earlier release of vbox, to perhaps isolate the issue >> further? Since there is no guarantee for Oracle Solaris binary >> compatibility, this is kind of an ambiguous zone here. >> >> /dale >> > > Dale, you are probably right, that is a tough pill to swallow though since I > have a lot of scripts/tools to manage vbox and it works really well on OmniOS > -- when it works :)
One of the underlying reasons for this is that illumos (and by extension, OmniOS) still identify as “SunOS 5.11”, which the vbox build system equates to “Oracle Solaris 11” Since Oracle Solaris 11 and illumos are moving on in their own directions and the technical differences between them only grow over time, this is a pain point when it comes to trying to get binaries that were compiled on one to work on the other. A good example of this in the context of vbox is vboxfs, which now uses some symbols added in a Oracle Solaris 11 update, but do not exist on illumos. /dale _______________________________________________ OmniOS-discuss mailing list OmniOS-discuss@lists.omniti.com http://lists.omniti.com/mailman/listinfo/omnios-discuss
