Crashme test

Crashme test

[Pedro Gracia]

Hi all,

I'm using Xen in a desktop enviroment with debian and fdl for about six 
weeks and finally, after a lot of weeks with continuous resets (instant 
shutdowns) while I'm using firefox in my debian dom0 and the same resets 
with a conexant usb modem (while the modem is starting the beeps for the 
call), I found a test that reset my Xen box in less of 1 second...

ii  crashme        2.4-5          Stress tests operating system stability

My box pass the crashme test with a normal 2.6.11.8 kernel for a few 
minutes (I didn't to use crashme for an hour).

Someone can test this? I'm using Xen 2.0.6 (stable) but I had the same 
resets with an Xen-unstable too.

Thanks a lot,

Pedro

Re: Crashme test

[Vincent Hanquez]

On Mon, May 23, 2005 at 11:56:29PM +0100, Pedro Gracia wrote:
> Hi all,
> 
> I'm using Xen in a desktop enviroment with debian and fdl for about six 
> weeks and finally, after a lot of weeks with continuous resets (instant 
> shutdowns) while I'm using firefox in my debian dom0 and the same resets 
> with a conexant usb modem (while the modem is starting the beeps for the 
> call), I found a test that reset my Xen box in less of 1 second...
> 
> ii  crashme        2.4-5          Stress tests operating system stability
> 
> My box pass the crashme test with a normal 2.6.11.8 kernel for a few 
> minutes (I didn't to use crashme for an hour).
> 
> Someone can test this? I'm using Xen 2.0.6 (stable) but I had the same 
> resets with an Xen-unstable too.

The crashme problem should now be resolved in testing and unstable.

It would be helpful if more people give Xen some more stress with crashme
just to be sure.

Thanks,
-- 
Vincent Hanquez

Re: Crashme test

[Pedro Gracia]

Hi. Thanks! Good job!

Test complete, total real time: 3613 seconds (0 01:00:13)
exit status ... number of cases
      26368 ...     1
      31744 ...     1
      56576 ...     1
          0 ...     1
      64000 ...     2
        256 ...     1
       3072 ...     1
      29696 ...     1
      49152 ...     1
      56064 ...     1
       8448 ...     1
       8960 ...     1
      65280 ...     1
          9 ...    50
      25600 ...   119
         11 ...  3548
 
Cheers,

Pedro

Vincent Hanquez wrote:

>On Mon, May 23, 2005 at 11:56:29PM +0100, Pedro Gracia wrote:
>  
>
>>Hi all,
>>
>>I'm using Xen in a desktop enviroment with debian and fdl for about six 
>>weeks and finally, after a lot of weeks with continuous resets (instant 
>>shutdowns) while I'm using firefox in my debian dom0 and the same resets 
>>with a conexant usb modem (while the modem is starting the beeps for the 
>>call), I found a test that reset my Xen box in less of 1 second...
>>
>>ii  crashme        2.4-5          Stress tests operating system stability
>>
>>My box pass the crashme test with a normal 2.6.11.8 kernel for a few 
>>minutes (I didn't to use crashme for an hour).
>>
>>Someone can test this? I'm using Xen 2.0.6 (stable) but I had the same 
>>resets with an Xen-unstable too.
>>    
>>
>
>The crashme problem should now be resolved in testing and unstable.
>
>It would be helpful if more people give Xen some more stress with crashme
>just to be sure.
>
>Thanks,
>  
>

RE: Crashme test

[Ian Pratt]

> I'm using Xen in a desktop enviroment with debian and fdl for 
> about six weeks and finally, after a lot of weeks with 
> continuous resets (instant
> shutdowns) while I'm using firefox in my debian dom0 and the 
> same resets with a conexant usb modem (while the modem is 
> starting the beeps for the call), I found a test that reset 
> my Xen box in less of 1 second...
> 
> ii  crashme        2.4-5          Stress tests operating 
> system stability
> 
> My box pass the crashme test with a normal 2.6.11.8 kernel 
> for a few minutes (I didn't to use crashme for an hour).
> 
> Someone can test this? I'm using Xen 2.0.6 (stable) but I had 
> the same resets with an Xen-unstable too.

Yep, I can reproduce this using 'crashme +2000 666 100 1:00:00' with the
binary at http://www.cl.cam.ac.uk/~iap10/temp/crashme

It happens in domU as well as dom0, which should make it easier to debug
as the test cycle is quicker. 


Ian