Re: [Qemu-devel] [PATCH v2 0/6] Mirrored writes using blockdev-transaction

[Anthony Liguori <anthony@codemonkey.ws>]

On 03/05/2012 02:53 AM, Kevin Wolf wrote:
> Am 01.03.2012 22:10, schrieb Anthony Liguori:
>> On 03/01/2012 05:21 AM, Paolo Bonzini wrote:
>>> This implements all ingredients to establish mirrored writes.
>>> The drive-reopen command that is used to terminate mirrored writes
>>> is not included in this series.
>>>
>>> Tested with the following scenarios:
>>>
>>> a) mirror only
>>>
>>> 1) create base.qcow2 and start QEMU with it
>>>
>>> 2) Execute the following QMP command
>>>
>>> { "execute": "qmp_capabilities" }
>>> { "execute": "blockdev-transaction", "arguments":
>>>     {'actions': [
>>>       { 'type': 'mirror', 'data' :
>>>         { 'device': 'ide0-hd0', 'target': '/home/pbonzini/mirror.qcow2' } } ] } }
>>> { "execute": "cont" }
>>>
>>> 3) hibernate the guest (this requires an IDE disk and -cpu kvm64,-kvmclock)
>>>
>>> 4) restart the guest with mirror.qcow2
>>>
>>>
>>> b) atomic snapshot+mirror
>>>
>>> 1) start QEMU with an existing image test.img
>>>
>>> 2) Execute the following QMP command
>>>
>>> { "execute": "qmp_capabilities" }
>>> { "execute": "blockdev-transaction", "arguments":
>>>     {'actions': [
>>>       { 'type': 'snapshot', 'data' :
>>>         { 'device': 'ide0-hd0', 'snapshot-file': '/home/pbonzini/base.qcow2' } },
>>>       { 'type': 'mirror', 'data' :
>>>         { 'device': 'ide0-hd0', 'target': '/home/pbonzini/mirror.qcow2' } } ] } }
>>> { "execute": "cont" }
>>
>> We don't have schema introspection today.  How would one determine when new
>> transaction types are available?
>>
>> I think we need some sort of introspection method too in order for clients to
>> figure out when the command is extended.
>
> How about coupling the types with independently available commands for
> now? We would rename 'snapshot' to 'blockdev-snapshot-sync', which does
> the same thing outside of transactions. The mirror patches would then
> introduce a 'drive-mirror' top-level command at the same time as they
> introduce a 'drive-mirror' transaction type.

I think it's reasonable but we need to make sure to document the relationship 
here explicitly in the schema.

Regards,

Anthony Liguori

>
> Kevin