Re: [Qemu-devel] [PATCH v1 1/1] qemu-doc: update gluster protocol usage guide

[Eric Blake <eblake@redhat.com>]

[-- Attachment #1: Type: text/plain, Size: 1794 bytes --]

On 11/02/2016 07:03 AM, Prasanna Kalever wrote:

>> "debug":"N" does not match the schema; the parameter is named
>> "debug-level", and it is an integer not a string.  The parameter is
>> optional; you could just omit it. But if you are going to include it,
>> give a reasonable example like "debug-level":0.
> 
> Eric,
> 
>>From the code I can see
> #define GLUSTER_OPT_DEBUG           "debug"
> 
> And even tried this, there is no option 'debug-level'.
> And yes, I should admit it is 'int' not a 'string'

Uggh. We were under so much pressure to try and get GlusterServer into
2.7 that we really botched it.

The QMP definition is:

{ 'struct': 'BlockdevOptionsGluster',
  'data': { 'volume': 'str',
            'path': 'str',
            'server': ['GlusterServer'],
            '*debug-level': 'int',
            '*logfile': 'str' } }

and what's more, we've copied that string elsewhere for 2.8:

{ 'struct': 'BlockdevOptionsNfs',
  'data': { 'server': 'NFSServer',
            'path': 'str',
            '*user': 'int',
            '*group': 'int',
            '*tcp-syn-count': 'int',
            '*readahead-size': 'int',
            '*page-cache-size': 'int',
            '*debug-level': 'int' } }

So either we need to fix the QMP to spell it 'debug' (to match the
documentation you gave); or fix the docs to match the QMP and add some
back-compat glue into the gluster code to manage both spellings rather
than just the QMP spelling.

Since blockdev-add is still experimental, I'm leaning towards changing
the QMP for both gluster and NFS.  But we need to get it done before
2.8.  Looks like I have some patches to propose.

-- 
Eric Blake   eblake redhat com    +1-919-301-3266
Libvirt virtualization library http://libvirt.org


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 604 bytes --]