[libvirt] [PATCH v3 0/3] qemu: add support for multiple gluster hosts/servers

[Peter Krempa]

On Fri, Jul 22, 2016 at 15:14:47 +0530, Prasanna Kalever wrote:
> On Fri, Jul 22, 2016 at 2:03 PM, Peter Krempa <pkrempa at redhat.com> wrote:
> > On Fri, Jul 22, 2016 at 13:50:28 +0530, Prasanna Kumar Kalever wrote:
> >> These series of patches are rebased on latest master;
> >> The dependent QEMU patches are now merged on master targeting for 2.7
> >
> > I actually started implementing this already. The precursor for this
> > series is the JSON backing store parser [1].
> 
> I remember you comment on my v2 series saying "we should wait until
> qemu design settles ?"
> So I have focused on them first, since they got merged on master
> yesterday, I have resumed this patches :)
> 
> >
> > Since the backing store parser you've posted here is incomplete and
> > lacking some changes that I've requested in the review of the last
> > version (mostly that it will fail to compile without gluster enabled) I
> > will post my own implementation of that.
> >
> > I'll try to adapt your series on top of the code that I've
> > posted earlier and a few refactors that I've not yet posted.
> 
> From your patches provided at [1] I have noticed,
> 
> +static int
> +virStorageSourceParseBackingJSONGluster(virStorageSourcePtr src,
> +                                        virJSONValuePtr json,
> 
> [...]
> 
> +
> +    /* gluster currently supports only URI syntax passed in as filename */
> +    virReportError(VIR_ERR_INVALID_ARG, "%s",
> +                   _("missing gluster URI in JSON backing volume definition"));
> +
> +    return -1;
> +}
> 
> So this just has the skeleton atleast for now,

Yep. I've posted that before the qemu code settled.

> I think here our v3 patches should fit-in and do their job of JSON
> parsing and formatting
> IMO we already have most of the gluster work done here.

I wrote the patch on top of that series adding the JSON parser for qemu
already yesterday.

> 
> Let me know when you are expecting [1] go in, I shall wait for your
> patch series.

I'll actually take some bits of your series and try to integrate it.

> And will be happy to help you with a rebase of gluster patches on top
> of your pseudo JSON code.

It won't be necessary as I've said since I've already done a different
patch for that.

I'm planing of taking parts of 2/3 and using some stuff since as DanPB
pointed out the command line format isn't really what we should use. I
already wrote a json to commandline formatter for memory hotplug so I
plan to fix it for use for disk commands too.

I'll cc you on the stuff.

Peter