[Libvirt-cim] [PATCHv2 0/3] Make Shutdown and Reboot state changes jobs

[Eduardo Lima (Etrunko)]

On 03/08/2012 10:07 AM, Eduardo Lima (Etrunko) wrote:
> On 03/07/2012 02:12 PM, Sharad Mishra wrote:
>> On Tue, 2012-03-06 at 14:55 -0300, Eduardo Lima (Etrunko) wrote:
>>> From: "Eduardo Lima (Etrunko)" <eblima at br.ibm.com>
>>>
>>> This code is based on the VSMigrationService for the job creation and on
>>> ComputerSystemIndication for the job thread, which will listen for the
>>> respective domain event to mark the job as finished.
>>>
>>> I am still not able to test this feature in my environment, every call to
>>> CBGetInstance fails with code error 100. This is happening in other places of
>>> the codebase, not only the ComputerSystem provider, but also ElementConfors,
>>> for instance. Please use the test case send in previous email for reference.
>>
>> Eduardo, It is not clear if you started seeing this error code after
>> applying this patch or were you getting it before.
> 
> It already happens with other test cases that will cause the provider to
> make a call such as CBGetInstance, CBCreateInstance and others from the
> same family. For example, test ElementConforms/02_reverse.py fails with
> the same error code 100, with or without this patch applied. I suspect
> it can be something messed up with my environment.
> 

Small update on this. I found the issue here with the error code 100. I
have dug deeper on tog-pegasus by enabling some debug messages and found
something interesting:

Http [19773:140483615913728:Socket.cpp:137]: connect() returned error
code 13

Looking in /usr/include/asm-generic/errno-base.h we get the following:

#define EACCES          13      /* Permission denied */

After disabling SELinux, further cimtest runs never returned the 'Error
100' anymore. I have worked a bit on the patch but it is still not working.

Now I'm stuck with a random crash happening on the state_change thread,
whenever calls to set properties on the job instance are placed. Looks
very likely to be an an issue on the tog-pegasus side, unless I am doing
something very wrong in this case.
-- 
Eduardo de Barros Lima
Software Engineer, Open Virtualization
Linux Technology Center - IBM/Brazil
eblima at br.ibm.com