qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Qemu-devel] [PATCH 13/18] blockdev: Rename I/O throttling options f


From: Kevin Wolf
Subject: Re: [Qemu-devel] [PATCH 13/18] blockdev: Rename I/O throttling options for QMP
Date: Fri, 26 Jul 2013 18:54:44 +0200
User-agent: Mutt/1.5.21 (2010-09-15)

Am 26.07.2013 um 18:44 hat Eric Blake geschrieben:
> On 07/26/2013 10:26 AM, Kevin Wolf wrote:
> 
> >> This patch will probably conflict with Benoît's work on leaky bucket
> >> throttling; can the two of you decide which one should go in first?  Are
> >> we trying to target both this series and leaky bucket throttling for 1.6?
> > 
> > If you complete the review before I leave today, I might still send a
> > pull request, but as I'm going to disable blockdev-add and the new
> > options once again for 1.6, it doesn't really matter that much.
> 
> In other words, just as it was in 1.5, the new parser is cool enough to
> implement the framework now to ease backport efforts, but untested
> enough that we'd rather defer use of that framework until after we are
> back out of freeze.

Right. Before actually committing to the interface, I'd like you to have
some real libvirt code running on it. I assume that's doable in the 1.7
time frame, right?

> > Benoît's series is for 1.7 as well, if I understood Stefan correctly. He
> 
> Even though the latest subject line requests for-1.6?
> https://lists.gnu.org/archive/html/qemu-devel/2013-07/msg04005.html

Yes, it was discussed on IRC, Benoît will target 1.7 now.

> > said he was going to merge a bug fix part of it for 1.6 and leave the
> > rest for 1.7. (I haven't been following the throttling series myself,
> > that's why I can't comment in much more detail.)
> 
> I guess I also need to comment on that series - we're late enough that
> bug fixes are okay, but new options are risky; and the tail end of that
> series adds new options to throttling as part of switching to a new
> algorithm.

Indeed, adding new options and switching the whole algorithm that late in
the cycle is something that I would find a little bit too scary.

Kevin



reply via email to

[Prev in Thread] Current Thread [Next in Thread]