[Kimchi-devel] [PATCH] Issue #999: Attach storage to guest on s390x without libvirt

Archana Singh archus at linux.vnet.ibm.com
Wed Sep 21 10:29:32 UTC 2016



On 09/20/2016 08:33 PM, Aline Manera wrote:
>
>
> On 09/20/2016 03:41 AM, Archana Singh wrote:
>>
>>
>>
>> On 09/19/2016 10:19 PM, Aline Manera wrote:
>>>
>>>
>>> On 09/19/2016 12:33 PM, Harshal Patil wrote:
>>>> Harshal Patil
>>>> E-mail: harshal.patil at in.ibm.com <mailto:harshal.patil at in.ibm.com>
>>>>
>>>>     ----- Original message -----
>>>>     From: Aline Manera <alinefm at linux.vnet.ibm.com>
>>>>     Sent by: kimchi-devel-bounces at ovirt.org
>>>>     To: harshalp at linux.vnet.ibm.com, Kimchi Devel
>>>>     <kimchi-devel at ovirt.org>
>>>>     Cc:
>>>>     Subject: Re: [Kimchi-devel] [PATCH] Issue #999: Attach storage
>>>>     to guest on s390x without libvirt
>>>>     Date: Sat, Sep 17, 2016 1:03 AM
>>>>
>>>>     On 09/12/2016 06:55 AM, harshalp at linux.vnet.ibm.com wrote:
>>>>     > From: Harshal Patil <harshalp at linux.vnet.ibm.com>
>>>>     >
>>>>     > This patch enables direct storage attachment to guests running
>>>>     > on s390x arch without going through libvirt
>>>>     >
>>>>     > Signed-off-by: Harshal Patil <harshalp at linux.vnet.ibm.com>
>>>>     > ---
>>>>     >   docs/API.md         |  3 +++
>>>>     >   i18n.py             |  3 +++
>>>>     >   model/vmstorages.py | 25 +++++++++++++++++++++++--
>>>>     >   utils.py            |  2 +-
>>>>     >   4 files changed, 30 insertions(+), 3 deletions(-)
>>>>     >
>>>>     > diff --git a/docs/API.md b/docs/API.md
>>>>     > index 6678ac5..c9d5d2a 100644
>>>>     > --- a/docs/API.md
>>>>     > +++ b/docs/API.md
>>>>     > @@ -239,6 +239,9 @@ Represents a snapshot of the Virtual
>>>>     Machine's primary monitor.
>>>>     >       * path: Path of cdrom iso.
>>>>     >       * pool: Storage pool which disk image file locate in.
>>>>     >       * vol: Storage volume name of disk image.
>>>>     > +    * dir_path: s390x specific attribute to attach direct
>>>>     storage without libvirt
>>>>     > +    * format: s390x specific attribute specify the format of
>>>>     direct storage
>>>>     > +    * size: s390x specific attribute to specify the size of
>>>>     direct storage
>>>>     >
>>>>     >   ### Sub-resource: storage
>>>>     >   **URI:** /plugins/kimchi/vms/*:name*/storages/*:dev*
>>>>     > diff --git a/i18n.py b/i18n.py
>>>>     > index fc7dbc7..ceb4286 100644
>>>>     > --- a/i18n.py
>>>>     > +++ b/i18n.py
>>>>     > @@ -332,6 +332,9 @@ messages = {
>>>>     >       "KCHVMSTOR0016E": _("Volume already in use by other
>>>>     virtual machine."),
>>>>     >       "KCHVMSTOR0017E": _("Only one of path or pool/volume
>>>>     can be specified to add a new virtual machine disk"),
>>>>     >       "KCHVMSTOR0018E": _("Volume chosen with format
>>>>     %(format)s does not fit in the storage type %(type)s"),
>>>>     > +    "KCHVMSTOR0019E": _("On s390x arch one of pool, path of
>>>>     dir_path must be specified"),
>>>>     > +    "KCHVMSTOR0020E": _("On s390x arch 'format' must be
>>>>     specified while attaching disk to virtual machine"),
>>>>     > +    "KCHVMSTOR0021E": _("Virtual disk already exists on the
>>>>     system: %(disk_path)s"),
>>>>     >
>>>>     >       "KCHSNAP0001E": _("Virtual machine '%(vm)s' must be
>>>>     stopped before creating a snapshot of it."),
>>>>     >       "KCHSNAP0002E": _("Unable to create snapshot '%(name)s'
>>>>     on virtual machine '%(vm)s'. Details: %(err)s"),
>>>>     > diff --git a/model/vmstorages.py b/model/vmstorages.py
>>>>     > index 2e9f783..156a16f 100644
>>>>     > --- a/model/vmstorages.py
>>>>     > +++ b/model/vmstorages.py
>>>>     > @@ -17,6 +17,7 @@
>>>>     >   # License along with this library; if not, write to the
>>>>     Free Software
>>>>     >   # Foundation, Inc., 51 Franklin Street, Fifth Floor,
>>>>     Boston, MA  02110-1301 USA
>>>>     >
>>>>     > +import os
>>>>     >   import string
>>>>     >   from lxml import etree
>>>>     >
>>>>     > @@ -30,6 +31,7 @@ from
>>>>     wok.plugins.kimchi.model.storagevolumes import StorageVolumeModel
>>>>     >   from wok.plugins.kimchi.model.utils import get_vm_config_flag
>>>>     >   from wok.plugins.kimchi.model.vms import DOM_STATE_MAP, VMModel
>>>>     >   from wok.plugins.kimchi.osinfo import lookup
>>>>     > +from wok.plugins.kimchi.utils import create_disk_image, is_s390x
>>>>     >   from wok.plugins.kimchi.xmlutils.disk import
>>>>     get_device_node, get_disk_xml
>>>>     >   from wok.plugins.kimchi.xmlutils.disk import
>>>>     get_vm_disk_info, get_vm_disks
>>>>     >
>>>>     > @@ -82,11 +84,30 @@ class VMStoragesModel(object):
>>>>     >           # Path will never be blank due to API.json
>>>>     verification.
>>>>     >           # There is no need to cover this case here.
>>>>     >           if not ('vol' in params) ^ ('path' in params):
>>>>     > -            raise InvalidParameter("KCHVMSTOR0017E")
>>>>     > +            if not is_s390x():
>>>>     > +                raise InvalidParameter("KCHVMSTOR0017E")
>>>>     > +
>>>>
>>>>     > +            if 'dir_path' not in params:
>>>>     > +                raise InvalidParameter("KCHVMSTOR0019E")
>>>>
>>>>     >The dir_path is exclusively for s390x. You need to add this
>>>>     verification
>>>>     >too.
>>>>
>>>> Not required. Check the condition again (The line which says 'if 
>>>> not is_s390x()'). Only if the platform is s390x it will enter the 
>>>> code path where 'dir_path' is checked so there is no need to check 
>>>> for s390x arch again. I have tested this flow on x86 as well as on 
>>>> s390x, and it works fine.
>>>
>>>
>>> Sorry but how the first condition is related to the second one?!
>>>
>>> If I am x86 and I have specified a 'vol' or a 'path' and not a 
>>> 'dir_path' as it is exclusively for s390x, it will raise the 
>>> dir_path exception (the second condition).
>>>
>>> So or you add a is_s390x() condition to the second statement or you 
>>> place the whole first block in a if conduction:
>>>
>>> if not is390x():
>>>      if not (vol in params) ^ (path in params):
>>>         (...)
>>> else:
>>>       if dir_path not in params:
>>>           raise ...
>>>
>> Below is the flow:
>>
>> 1) On x86, if path or volume is not present in params then raise 
>> Invalid parameter(KCHVMSTOR0017E). (Same as it is in upstream code)
>> 2) On x86 or s390x, if any one of them is present don't raise invalid 
>> parameter(KCHVMSTOR0017E).. (Same as it is in upstream code)
>> 3) On s390x, if path or volume is not present instead of raising 
>> invalid parameter, it should go and check for dir_path and if 
>> dir_path is also not present then only raise Invalid 
>> parameter(KCHVMSTOR0019E).
>>
>> So on x86, if I have specified a vol or a path then it will not go 
>> inside first if itself(not ('vol' in params) ^ ('path' in params) 
>> will be false).
>> and on x86, if I have neither vol nor path then it will go inside 
>> first if and also second one and raise invalid parameter.
>>
>> And on x86, due to first nested 'if' (that is not is_s390x()) it will 
>> never go to second nested 'if' ('dir_path' not in params).
>>
>> In Summary, on x86 if I have neither vol nor path then it should 
>> raise exception KCHVMSTOR0017E (same as currently upstream code 
>> behaves) and does not proceed further.
>> and on s390x we should check for vol and path, and if non of it is 
>> present then instead of raising exception KCHVMSTOR0017E go ahead and 
>> check for addition key dir_path. If dir_path is also not present then 
>> only raise exception KCHVMSTOR0019E.
>>
>> Hope I make sense to you.
>
> Yeap. That makes sense.
>
> But what about on x86 where I have path or vol and nor dir_path? How 
> do you prevent me to go to the second if and raise an exception?
>
> The second if is not nested as you said above.

On x86, check 'not is_s390x()' will always true and exception will be 
raised and hence it will never go further to check "'dir_path' not in 
params'".

>
>>
>>>>
>>>>
>>>>     >           dom = VMModel.get_vm(vm_name, self.conn)
>>>>     >           params['bus'] = _get_device_bus(params['type'], dom)
>>>>     > -        params['format'] = 'raw'
>>>>     > +
>>>>     > +        if is_s390x() and params['type'] == 'disk':
>>>>     > +            if 'format' not in params:
>>>>     > +                raise InvalidParameter("KCHVMSTOR0020E")
>>>>     > +            if 'dir_path' in params:
>>>>     > +                size = params['size']
>>>>     > +                name = params['name']
>>>>     > +                dir_path = params.get('dir_path')
>>>>     > +                params['path'] = dir_path + "/" + name
>>>>     > +                if os.path.exists(params['path']):
>>>>     > +                    raise InvalidParameter("KCHVMSTOR0021E",
>>>>     > + {'disk_path': params['path']})
>>>>     > +  create_disk_image(format_type=params['format'],
>>>>     > +  path=params['path'], capacity=size)
>>>>     > +        else:
>>>>     > +            params['format'] = 'raw'
>>>>     >
>>>>     >           dev_list = [dev for dev, bus in
>>>>     get_vm_disks(dom).iteritems()
>>>>     >                       if bus == params['bus']]
>>>>     > diff --git a/utils.py b/utils.py
>>>>     > index 0fca191..7fdec02 100644
>>>>     > --- a/utils.py
>>>>     > +++ b/utils.py
>>>>     > @@ -305,4 +305,4 @@ def create_disk_image(format_type, path,
>>>>     capacity):
>>>>     >       if rc != 0:
>>>>     >           raise OperationFailed("KCHTMPL0035E", {'err': err})
>>>>     >
>>>>     > -    return
>>>>     > \ No newline at end of file
>>>>     > +    return
>>>>
>>>>     _______________________________________________
>>>>     Kimchi-devel mailing list
>>>>     Kimchi-devel at ovirt.org
>>>>     http://lists.ovirt.org/mailman/listinfo/kimchi-devel
>>>>
>>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> Kimchi-devel mailing list
>>> Kimchi-devel at ovirt.org
>>> http://lists.ovirt.org/mailman/listinfo/kimchi-devel
>>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ovirt.org/pipermail/kimchi-devel/attachments/20160921/a684ebdc/attachment.html>


More information about the Kimchi-devel mailing list