[vbox-dev] Fw: Setting HPET via SOAP

Klaus Espenlaub klaus.espenlaub at oracle.com
Thu Nov 27 10:32:16 UTC 2014

On 27.11.2014 10:26, Yuriy Tretyakov wrote:
> We need to run our code under python 3.1 that's why we need to use 
> suds here.
> Anyway we may face this situation without python at all. I ran SOAP 
> UI. I ran a list of command-one by one.
Ah, python 3.x support for the ZSI bindings is on the todo list, 
unfortunately as always with far more work than developers...
> I lock the machine and run ISession_getMachine. webserver returned new 
> mutable machine ID.
> Runing ISession_getType for my session id - I got
> <vbox:ISession_getTypeResponse>
>          <returnval>WriteLock</returnval>
>       </vbox:ISession_getTypeResponse>
> So machine is in mutable state now, right?
Yes and no. When there is a properly established session there are two 
IMachine objects for every "VM", one readonly (the one you'd get when 
digging out the IMachine object when there is no session) and one 
mutable. You're definitely using the right one, otherwise you wouldn't 
get the error message below. So that's good news, your code for 
establishing the session etc. is correct.
> Runing IMachine_setHPETEnabled  for mutable ID
> <_this>5f0fac6e7e0179b0-0000000000000011</_this>
>          <HPETEnabled>true</HPETEnabled>
> !
> got error VirtualBox error: rc=0x80bb0002 The machine is not mutable 
> (state is Saved) (0x80bb0002)</faultstring>
Ah, there you go. I asked you already what kind of snapshot you're 
restoring - and it's clearly an online snapshot. In such a case you 
can't change the settings, as the VM isn't powered off. So as I 
suspected a few mails ago, your assumptions are wrong.

To repeat: a VM in state "Saved" can't be significantly reconfigured, 
because it is logically still running, with the entire hardware state 
saved (a bit like a hibernated PC, where one also has to keep fingers 
off the hardware config). It's simply not possible to enable HPET in 
this situation. It would be if you throw away the saved state, but I'm 
not sure if that's making the overall situation worse.

So what you're seeing is 100% expected and correct behavior. The only 
mystery is why the memory size can be changed, because that shouldn't be 
possible either.
> Just for your information. It seems to me a little strange that when 
> vbox server returning an error it used not the same value in returnval 
> as my mutable machine id. It increased each time request being sent to 
> server. Here is what SOAP ui shows:
>  <faultcode>SOAP-ENV:Client</faultcode>
>          <faultstring>VirtualBox error: rc=0x80bb0002 The machine is 
> not mutable (state is Saved) (0x80bb0002)</faultstring>
>          <detail>
>             <vbox:RuntimeFault>
>  <resultCode>-2135228414</resultCode>
>           &! nbsp; 
>  <returnval>5f0fac6e7e0179b0-0000000000000033</returnval>
>             </vbox:RuntimeFault>
>          </detail>
>       </SOAP-ENV:Fault>
I don't follow - the returned reference is not for IMachine, it is an 
error object (IVirtualBoxError) which describes the error in more detail 
than what the SOAP fault code can do. Each fault creates a new error 
object, which explains why a new reference is created.

Phew, at least the biggest mystery is resolved.

> --- Исходное сообщение ---
> От кого: "Klaus Espenlaub" <klaus.espenlaub at oracle.com>
> Дата: 26 ноября 2014, 17:53:21
>     On 26.11.2014 16:09, Yuriy Tretyakov wrote:
>>     Hi Klaus.
>>     Thank you for your quick answer!
>>     Sure, vm machine is turned off. I trying to turn HPET on after
>>     snapshot is restored (machine is down) but got this error.
>>     And yes , as I understand from SDK, such settings shall work with
>>     mutable machine. Setting RAM size works fine, but HPET failed.
>>     I specify that  IMachine_setCPUHotPlugEnabled doesn't work this
>>     way. Few month ago we were try to set videocapturing on and also
>>     no positive results. Could it be a defect with processing boolean
>>     values when we work with vboxwebsrv?
>     Can't tell from here what's "going over the wire" with your code,
>     but as long as the error message has the word "mutable" in it I
>     wouldn't expect trouble with the parameters as such, more with the
>     references.
>     Is there a particular reason why you can't/don't want to use ZSI?
>     It's what e.g. vboxshell.py uses if one selects the webservice. It
>     gives a nice, direct python object interface to the API, and can
>     also talk to the local API when it runs on the same system...
>     Klaus
>>     Regards,
>>     Yuriy
>>     --- Исходное сообщение ---
>>     От кого: "Klaus Espenlaub" <klaus.espenlaub at oracle.com>
>>     <mailto:klaus.espenlaub at oracle.com>
>>     Дата: 26 ноября 2014, 17:02:12
>>         Hi Yuriy,
>>         On 26.11.2014 15:42, Yuriy Tretyakov wrote:
>>>             Hi Guys!
>>>             Forum  moderator suggests me to contact somebody from
>>>             developers to solve my problem faster than waiting
>>>             answer on forum.
>>>             I have a trouble setting boolean values via SOAP for VBOX.
>>>             Actually, I need to set IMachine_setHPETEnabled to True
>>>             programmatically, but IMachine_setCPUHotPlugEnabled also
>>>             doesn't work.
>>>             Trying to set these parameters always follows with
>>>             Machine is not mutable error. But setting of memory like
>>>             in SDK example works fine.
>>>             What I'm actually do. I using suds SOAP client library
>>>             for Python.
>>         This doesn't sound like a problem with specific bindings to
>>         me (never used suds though, only used ZSI as you can see in
>>         the samples in the SDK). From the symptoms you're using the
>>         wrong IMachine reference. On the other hand, I can't explain
>>         why setting the memory size with the same reference works,
>>         it's doing the exact same checks on the API implementation side.
>>         Are you sure that the snapshot you are restoring is an
>>         "offline" one, i.e. for a powered off VM? Otherwise the VM is
>>         in saved state, which means no significant reconfiguration is
>>         possible (but again, same rules for HPET and memory size...).
>>         Klaus 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.virtualbox.org/pipermail/vbox-dev/attachments/20141127/069e5d7e/attachment-0001.html 

More information about the vbox-dev mailing list