[OmniOS-discuss] [discuss] Kernel Panic OmniOS r151006 svc.configd

[Dan McDonald]

> On Mar 16, 2015, at 4:02 PM, takashi ary <takashiary at gmail.com> wrote:
> 
>> Normally I like to see the dumps themselves, but these are both the same panic, and both in a seemingly innocent return from a door call.
> 
> It's possible to send vmdump.
> What is good way to send?

Given what you say below, I don't think you will need to send me anything...

>> Both panics are page faults, like the kernel was using a userspace pointer or something.  I don't know the doorfs subsystem that well, but given I've not seen this anywhere else, I'm wondering if something odd is going on inside VMware's memory management (you did say this is running inside VMware)?
> 
> This is running inside VMware ESXi 5.1 Update 2
> so search the VMware Knowledge Base...
> 
> Windows 2008 R2, Red Hat Enterprise Linux and Solaris 10 64-bit
> virtual machines blue screen or kernel panic when running on ESXi 5.x
> with an Intel E5/E7/E3 v2 series processor (2073791)
> http://kb.vmware.com/kb/2073791
> 
> $ prtconf -v | grep Xeon
>                        value='Intel(r) Xeon(r) CPU E5-2697 v2 @ 2.70GHz'
>                        value='Intel(r) Xeon(r) CPU E5-2697 v2 @ 2.70GHz'
> 
> Intel E5 v2 series processor!
> Bingo?

Yep!  See if the VMware note has information on the coredump for Solaris 10 --> it'll be close to any illumos distro, including OmniOS.  According to the note, you need "update 3" to 5.1.

>> No specific activity was going on prior to the panics, right?
> 
> Right, I think no one was using the server at that time.
> 
> This info may be better shared on the illumos developer list?

Yes, INCLUDING the VMware technical note and its solution.  Please share!!!  I'm glad this isn't a problem with us, but with VMware.  One major user of illumos runs on VMware, and needs to see that, but I suspect they know it already.

Thanks!
Dan