libvirt users - Oct 2015 - Re: "failed to connect to the hypervisor"

On 2015-10-20 05:33, Martin Kletzander wrote:
> On Mon, Oct 19, 2015 at 03:57:39PM -0400, Ken D'Ambrosio wrote:
>> Hi, all.  Ubuntu host, attempting to get virsh working with
>> VirtualBox... and failing.  Here's what happens:
>> 
>> root@foobox:~# virsh -c vbox:///session list
>> error: failed to connect to the hypervisor
>> error: internal error: unable to initialize VirtualBox driver API
>> 
> 
> What version of VirtualBox do you have installed?  The logging in that
> driver is not particularly loud, so this won't tell you much more than
> the fact that the initialization just failed.  It's most probably
> something missing in your system or you have to new version installed.
You're on the money: I do have the latest and greatest (5.0.6) from 
Oracle installed.  Despite some of its nifty features, I'm not wedded to 
it, if downgrading will fix my underlying issue.  Is that the 
recommended course of action?  Or is there something else I could do?

Thanks much!

-Ken

>> ------------------------------------------------------
>> 
>> Now, most of the docs I read about "failed to connect to the
>> hypervisor" have other errors that happen *before* the
"failed to
>> connect" one, that are the root cause (e.g., unable to open
socket, or
>> bad cert, or whatever).  But that's not the case, here.  And what
>> puzzles me more is that -- while I don't know what the output
*should*
>> look like -- to my eye, strace seems to say that we're getting
>> information back:
>> 
>> 1940  read(5, "Node 1 MemTotal:       33011680 kB\nNode 1 MemFree:
>> 32565612 kB\nNode 1 MemUsed:          446068 kB\nNode 1 Active:
>> 57948 kB\nNode 1 Inactive:          77036 kB\nNode 1 Active(anon):
>> 27688 kB\nNode 1 Inactive(anon):       84 kB\nNode 1 Active(file):
>> 30260 kB\nNode 1 Inactive(file):    76952 kB\nNode 1 Unevictable:
>> 52 kB\nNode 1
>> [...]
>> 
>> ------------------------------------------------------
>> 
>> In my googling, I see mention of per-VM XML files that... do
>> something.  But what they do, and how they're generated,
doesn't seem
>> clear.  (E.g., XML mentioned here: http://libvirt.org/drvvbox.html --
>> it just sort of sits there, with no context.)  Is that something I
>> need?
>> 
>> Thanks for any enlightenment you might be able to offer me,
>> 
>> -Ken
>> 
>> _______________________________________________
>> libvirt-users mailing list
>> libvirt-users@redhat.com
>> https://www.redhat.com/mailman/listinfo/libvirt-users

On Tue, Oct 20, 2015 at 11:59:03AM -0400, Ken D'Ambrosio
wrote:
>On 2015-10-20 05:33, Martin Kletzander wrote:
>>On Mon, Oct 19, 2015 at 03:57:39PM -0400, Ken D'Ambrosio wrote:
>>>Hi, all.  Ubuntu host, attempting to get virsh working with
>>>VirtualBox... and failing.  Here's what happens:
>>>
>>>root@foobox:~# virsh -c vbox:///session list
>>>error: failed to connect to the hypervisor
>>>error: internal error: unable to initialize VirtualBox driver API
>>>
>>
>>What version of VirtualBox do you have installed?  The logging in that
>>driver is not particularly loud, so this won't tell you much more
than
>>the fact that the initialization just failed.  It's most probably
>>something missing in your system or you have to new version installed.
>
>You're on the money: I do have the latest and greatest (5.0.6) from
>Oracle installed.  Despite some of its nifty features, I'm not wedded
>to it, if downgrading will fix my underlying issue.  Is that the
>recommended course of action?  Or is there something else I could do?
>
Well, of course the best thing would be adding support for it into
libvirt =)  If it is backwards compatible with the latest version that
has support in libvirt, then it shouldn't be a problem, I think, but I
have very limited experience with the vbox driver in libvirt.  Feel
free to have a look at the code and let me know, I'm sure we'll come
up with something.  In the meantime I'll compile the 5.0.8 version (so
I don't have to agree with the PUEL license) and see what I can do.
>Thanks much!
>
>-Ken
>
>
>>>------------------------------------------------------
>>>
>>>Now, most of the docs I read about "failed to connect to the
>>>hypervisor" have other errors that happen *before* the
"failed to
>>>connect" one, that are the root cause (e.g., unable to open
socket, or
>>>bad cert, or whatever).  But that's not the case, here.  And
what
>>>puzzles me more is that -- while I don't know what the output
*should*
>>>look like -- to my eye, strace seems to say that we're getting
>>>information back:
>>>
>>>1940  read(5, "Node 1 MemTotal:       33011680 kB\nNode 1
MemFree:
>>>32565612 kB\nNode 1 MemUsed:          446068 kB\nNode 1 Active:
>>>57948 kB\nNode 1 Inactive:          77036 kB\nNode 1 Active(anon):
>>>27688 kB\nNode 1 Inactive(anon):       84 kB\nNode 1 Active(file):
>>>30260 kB\nNode 1 Inactive(file):    76952 kB\nNode 1 Unevictable:
>>>52 kB\nNode 1
>>>[...]
>>>
>>>------------------------------------------------------
>>>
>>>In my googling, I see mention of per-VM XML files that... do
>>>something.  But what they do, and how they're generated,
doesn't seem
>>>clear.  (E.g., XML mentioned here: http://libvirt.org/drvvbox.html
--
>>>it just sort of sits there, with no context.)  Is that something I
>>>need?
>>>
>>>Thanks for any enlightenment you might be able to offer me,
>>>
>>>-Ken
>>>
>>>_______________________________________________
>>>libvirt-users mailing list
>>>libvirt-users@redhat.com
>>>https://www.redhat.com/mailman/listinfo/libvirt-users
>

On Wed, Oct 21, 2015 at 08:41:20AM +0200, Martin Kletzander
wrote:
>On Tue, Oct 20, 2015 at 11:59:03AM -0400, Ken D'Ambrosio wrote:
>>On 2015-10-20 05:33, Martin Kletzander wrote:
>>>On Mon, Oct 19, 2015 at 03:57:39PM -0400, Ken D'Ambrosio wrote:
>>>>Hi, all.  Ubuntu host, attempting to get virsh working with
>>>>VirtualBox... and failing.  Here's what happens:
>>>>
>>>>root@foobox:~# virsh -c vbox:///session list
>>>>error: failed to connect to the hypervisor
>>>>error: internal error: unable to initialize VirtualBox driver
API
>>>>
>>>
>>>What version of VirtualBox do you have installed?  The logging in
that
>>>driver is not particularly loud, so this won't tell you much
more than
>>>the fact that the initialization just failed.  It's most
probably
>>>something missing in your system or you have to new version
installed.
>>
>>You're on the money: I do have the latest and greatest (5.0.6) from
>>Oracle installed.  Despite some of its nifty features, I'm not
wedded
>>to it, if downgrading will fix my underlying issue.  Is that the
>>recommended course of action?  Or is there something else I could do?
>>
>
>Well, of course the best thing would be adding support for it into
>libvirt =)  If it is backwards compatible with the latest version that
>has support in libvirt, then it shouldn't be a problem, I think, but I
>have very limited experience with the vbox driver in libvirt.  Feel
>free to have a look at the code and let me know, I'm sure we'll come
>up with something.  In the meantime I'll compile the 5.0.8 version (so
>I don't have to agree with the PUEL license) and see what I can do.
>
So I went ahead, compiled the 5.0.8, installed it, modified libvirt to
use the 5.0.8 the same way as it already does use 4.3.4 and, of
course, they are not compatible.  I then spent some time looking at
the code and trying to find out what needs to be changed, but it
turned out the code is not easy to follow for someone who hadn't
looked at it.  I don't have enough time to work on that, but it might
be a good idea for someone to implement.  And the patches would be
very welcome, I believe =)

Have a nice day,
Martin