[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-block] [PATCH v4 1/2] live-block-ops.txt: Rename, rewrite, and
Re: [Qemu-block] [PATCH v4 1/2] live-block-ops.txt: Rename, rewrite, and improve it
Tue, 4 Jul 2017 12:08:33 +0200
On Thu, Jun 29, 2017 at 02:28:55PM +0200, Alberto Garcia wrote:
> On Wed 28 Jun 2017 04:58:00 PM CEST, Kashyap Chamarthy wrote:
> > +Once a 'mirror' job has started, there are two possible actions when a
> > +``drive-mirror`` job is active:
> > +
> > +1. Issuing the command ``block-job-cancel``: will, after completing
> > + synchronization of the content from the disk image chain to the
> > + target image, [E] -- create a point-in-time (which is at the time of
> > + *triggering* the cancel command) copy, contained in image [E], of the
> > + backing file.
> A point-in-time [...] copy [...] of the backing file ? That would be a
> copy of the whole chain or the topmost image (depending on the sync
Yep you're right -- it indeed copies the whole chain. I'll fix the
Kevin also made a good point on IRC (#qemu, OFTC): "By the way, are you
sure that block-job-cancel for mirror creates a point-in-time copy when
the cancel command is issued? Shouldn't it really be when the cancel
event is emitted?"
So, I'll spell out the mention of the event.
> > +The ``"sync": "full"``, from the above, means: copy the *entire* chain
> > +to the destination.
> I think it's in general a good idea to describe the different sync modes
> before doing this, because they're not completely obvious when you first
> try to use these commands.
> > +Notes on ``blockdev-mirror``
> > +~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> > +(1) Create the target image (using ``qemu-img``), say, backup.qcow2
> If we're doing a mirror here, backup.qcow2 is probably not the best
> name :-)
Yeah, I'll use the 'mirror.qcow2' unless someone has a better name :-)
I actually _did_ consider naming, but stuck with "backup" because,
`mirror` is a _kind_ of a backup, too. And "mirror" itself is a
slightly wrong name, as Stefan Hajnoczi once correcte me elsewhere: it's
synchronizing disk contents to the target, but not "live mirroring"
> > +Live disk backup --- ``drive-backup`` and ``blockdev-backup``
> > +-------------------------------------------------------------
> > +
> > +The ``drive-backup`` (and its newer equivalent ``blockdev-backup``) allows
> > +you to create a point-in-time snapshot.
> > +
> > +In this case, the point-in-time is when you *start* the ``drive-backup``
> > +(or its newer equivalent ``blockdev-backup``) command.
> > +
> > +Currently, there are four different types of synchronization modes:
> > +
> > +(1) ``full`` -- Synchronize the content of entire disk image chain to
> > + the target
> > +(2) ``top`` -- Synchronize only the contents of the top-most disk image
> > + in the chain to the target
> > +(3) ``none`` -- Synchronize only the new writes from this point on
> > +(4) ``incremental`` -- Synchronize content that is described by the
> > + dirty bitmap
> Here's the kind of summary of the different sync modes that I was
> talking about earlier.
I will move it to the top. I even thought about it, but forgot to act
on it. It makes logical sense to mention the terms on their first
> One note about sync=none, though. If I'm not wrong drive-backup makes a
> point-in-time copy of the data at the moment when the command is issued
> (that's one difference from drive-mirror). Therefore drive-backup never
> synchronizes "new writes from this point on".
> I don't think that drive-backup sync=none copies any data then, can
> anyone clarify?
Kevin clarified on IRC:
[kashyap] kwolf: When you get a moment, can you clarify Alberto's
question on this thread:
[kwolf] kashyap: The documentation [I think he means QAPI schema
documentation here] isn't very clear, we should probably improve
the comment in the schema
[kwolf] But looking at the code it seems it just disables the background
copy, i.e. it backups only what is overwritten.
[kashyap] kwolf: Let's see if I get it: so it ('drive-backup sync=none')
backups whatever the overwritten contents in the disk image
chain to the target?
[kwolf] kashyap: Yes. Normally backup consists of two parts: Anything
that is overwritten by the guest is first copied out to the
backup, and in the background the whole image is copied start to
end. With sync=none, it's only the first part.
> The rest of the document looks good to me, thanks!
Thanks for the review. Much appreciated.
|[Prev in Thread]
||[Next in Thread]|
- Re: [Qemu-block] [PATCH v4 1/2] live-block-ops.txt: Rename, rewrite, and improve it,
Kashyap Chamarthy <=