[libvirt] [PATCH v2] Add new attribute writeout to <filesystem> element

Deepak C Shetty deepakcs at linux.vnet.ibm.com
Mon Dec 26 12:29:21 UTC 2011

On 12/22/2011 11:07 PM, Daniel P. Berrange wrote:
> On Thu, Dec 22, 2011 at 05:25:49PM +0100, Paolo Bonzini wrote:
>> On 12/22/2011 03:20 PM, Eric Blake wrote:
>>>>>   Usage:
>>>>>   <filesystem type='mount' accessmode='passthrough' writeout='immediate'>
>>>>>      <source dir='/export/to/guest'/>
>>>>>      <target dir='mount_tag'/>
>>>>>   </filesystem>
>>> I still don't like the proposed XML.  We need to get consensus on where
>>> to put it, with my proposal being:
>>> <filesystem type='mount'>
>>>    <driver type='path' write='immediate'/>
>>>    <source dir='/export/to/guest'/>
>>>    <target dir='mount_tag'/>
>>> </filesystem>
>> Right, other types of drivers might include a userspace NFS client,
>> or some kind of "artificial" filesystem.
>> In addition, instead of write='immediate' I suggest
>> cache='writeback'/cache='writethrough'.
> Yeah I agree - we should follow the syntax we already use for caching
> with the<disk>  element instead of inventing new syntax
> Daniel
Hi Dan and Paolo,
      There had been a discussion earlier on the qemu-devel mailing list 
on whether
cache=writethrough is appropriate for 9pfs Vs writeout as the attribute 
and it was decided to use writeout as the writeout behaviour is not 
exactly what
cache=writethrough means. Ccing Aneesh here who was part of the original
qemu-devel discussion.


     2 things being discussed here from 9pfs perspective
1) Whether writeout should be an attribute at the <filesystem...> level 
or <driver..> level
in the XML schema.
2) Have cache=writethrough in the XML instead of writeout=immediate.

Comments pls.

More information about the libvir-list mailing list