Re: [PATCH] KVM-test: Add a ENOSPC subtest

[Kevin Wolf <kwolf@redhat.com>]

Am 14.01.2011 06:28, schrieb Amos Kong:
> ----- Original Message -----
>> From: Amos Kong <akong@redhat.com>
>>
>> KVM guest always pauses on NOSPACE error, this test
>> just repeatedly extend guest disk space and resume guest
>> from paused status.
>>
>> Changes from v2:
>> - Oops! Forgot to update tests_base.cfg.sample
>>
>> Changes from v1:
>> - Use the most current KVM test API
>> - Use the autotest API for external commands execution
>> - Instead of chaining multiple shell commands as pre and
>> post commands, create proper pre and post scripts for the
>> test, as it is easier to figure out problems
>> - Instead of setting up /dev/loop0 hardcoded by default,
>> find the first available loop device before and use it.
>>
>> Signed-off-by: Amos Kong <akong@redhat.com>
> 
> Thank you, Lucas
> I've retested this patch, it's ok.
> 
> BTW, hi Kevin,  when I check the images during this test, got the following output, is it harmful?
> 
> Leaked cluster 60796 refcount=1 reference=0
> Leaked cluster 60797 refcount=1 reference=0
> Leaked cluster 60798 refcount=1 reference=0
> Leaked cluster 60799 refcount=1 reference=0
> Leaked cluster 60800 refcount=1 reference=0
> Leaked cluster 60801 refcount=1 reference=0
> Leaked cluster 60802 refcount=1 reference=0
> Leaked cluster 60803 refcount=1 reference=0
> Leaked cluster 60804 refcount=1 reference=0
> Leaked cluster 60805 refcount=1 reference=0
> Leaked cluster 60806 refcount=1 reference=0
> Leaked cluster 60807 refcount=1 reference=0
> Leaked cluster 63982 refcount=1 reference=0
> Leaked cluster 63983 refcount=1 reference=0
> Leaked cluster 63984 refcount=1 reference=0
> Leaked cluster 63985 refcount=1 reference=0
> Leaked cluster 63986 refcount=1 reference=0
> Leaked cluster 63987 refcount=1 reference=0
> Leaked cluster 63988 refcount=1 reference=0
> Leaked cluster 63989 refcount=1 reference=0
> Leaked cluster 63990 refcount=1 reference=0
> Leaked cluster 63991 refcount=1 reference=0
> Leaked cluster 63992 refcount=1 reference=0
> Leaked cluster 63993 refcount=1 reference=0
> Leaked cluster 63994 refcount=1 reference=0
> Leaked cluster 63995 refcount=1 reference=0
> Leaked cluster 63996 refcount=1 reference=0
> Leaked cluster 63997 refcount=1 reference=0
> Leaked cluster 63998 refcount=1 reference=0
> Leaked cluster 63999 refcount=1 reference=0
> 
> 867 leaked clusters were found on the image.
> This means waste of disk space, but no harm to data.

I suppose the last two lines of the output answer your question. ;-)

With I/O errors or qemu/host crashes, cluster leaks are fully expected.

Kevin