I forgot to add the mailing list.

---------- Forwarded message ----------
From: Ruijin Zhou <[email protected]>
Date: Wed, Jan 11, 2012 at 5:30 PM
Subject: Re: [Xen-API] Question about CrossPoolMigration-v3
To: Dave Scott <[email protected]>


Thank you for the information.

I have implemented a new command (storage_migrate) in tap-ctl before.
And that command can call the image copy function I wrote in
blktap2/drivers. However, I don't know how to translate the dom_id
into tap_id or pid. I read through the blktap2, but I cannot find a
function to do that. That is why I am thinking maybe I need the
dom_info from libxl.

In my understanding, the tapdisk does not have any information about
the domains. Thus, if we want to use command like tap-ctl
storage_migrate -v dom_id -n /new/path/to/storage, I think we have to
bother the libxl.

BTW, I found that when I do xl list -l. I can see the dom_id for tap
device is always 0. I don't know why.

Would you please share you opinion? I would appreciate that.

Thank you very much.

Best,
Ruijin





On Wed, Jan 11, 2012 at 4:30 PM, Dave Scott <[email protected]> wrote:
> Hi,
>
> (I've re-added the mailing list)
>
> Ruijin Zhou wrote:
>> Thank you Dave. This is very helpful.
>>
>> I think my goal is simpler. I just want to migrate the storage without
>> moving the memory and VCPU info. In other words, I don't need to
>> bother the original live migration code.
>>
>> In my case, I just need to move the disk image file of one VM from one
>> directory (SAN1) to new directory (SAN2). Then, I change the disk
>> handler (in libxl library) of that VM to points to the new image file.
>> I will also handle the dirty block during the image moving in block
>> drivers (block-aio.c).
>>
>> I don't know whether I miss anything or not. Would you please share
>> some opinions?
>
> That sounds ok to me. If you use blktap2 to talk to the storage, then you 
> will only have to tell xl/libxl the path to the generic blktap2 block device. 
> You can use the 'tap-ctl' command to redirect the blktap2 device to your new 
> image file, without having to tell xl/libxl anything.
>
> Cheers,
> Dave



--
With warm regards,

University of Florida,
Ruijin Zhou
[email protected]


-- 
With warm regards,

University of Florida,
Ruijin Zhou
[email protected]

_______________________________________________
xen-api mailing list
[email protected]
http://lists.xensource.com/mailman/listinfo/xen-api

Reply via email to