Re: [Qemu-devel] Re: [PATCH] Add cache=volatile parameter to -drive

[Anthony Liguori <anthony@codemonkey.ws>]

On 05/26/2010 03:52 AM, Aurelien Jarno wrote:
> On Tue, May 25, 2010 at 08:31:20PM -0500, Anthony Liguori wrote:
>    
>> On 05/25/2010 04:01 PM, Aurelien Jarno wrote:
>>      
>>> I really think this patch can be useful, in my own case when testing
>>> debian-installer (I already cache=writeback). In short all that is about
>>> developing and testing, as opposed to run a VM in production, can
>>> benefit about that. This was one of the original use case of QEMU before
>>> KVM arrived.
>>>
>>> Unless someone can convince me not to do it, I seriously considering
>>> applying this patch.
>>>        
>> There really needs to be an indication in the --help output of what
>> the ramifications of this option are, in the very least.  It should
>>      
> That's indeed something than can be done, but to avoid double standards,
> it should also be done for other features that can lead to data
> corruption. I am talking for example on the qcow format, which is not
> really supported anymore.
>    

I agree.

>> also be removable via a ./configure option because no sane
>> distribution should enable this for end users.
>>
>>      
> I totally disagree. All the examples I have given apply to qemu *users*,
> not qemu developers. They surely don't want to recompile qemu for their
> usage. Note also that what is proposed in this patch was the default not
> too long ago, and that a lot of users complained about the new default
> for their usage, they see it as a regression. We even had to put a note
> explaining that in the Debian package to avoid to many bug reports.
> cache=writeback only answer partially to this use case.
>    

It's hard for me to consider this a performance regression because 
ultimately, you're getting greater than bare metal performance (because 
of extremely aggressive caching).  It might be a regression from the 
previous performance, but that was at the cost of safety.

We might get 100 bug reports about this "regression" but they concern 
much less than 1 bug report of image corruption because of power 
failure/host crash.  A reputation of being unsafe is very difficult to 
get rid of and is something that I hear concerns about frequently.

I'm not suggesting that the compile option should be disabled by default 
upstream.  But the option should be there for distributions because I 
hope that any enterprise distro disables it.

Regards,

Anthony Liguori