[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Qemu-devel] [PATCH RFC] block/gluster: limit the transfer size to 5
From: |
Kevin Wolf |
Subject: |
Re: [Qemu-devel] [PATCH RFC] block/gluster: limit the transfer size to 512 MiB |
Date: |
Thu, 28 Mar 2019 16:42:35 +0100 |
User-agent: |
Mutt/1.11.3 (2019-02-01) |
Am 28.03.2019 um 16:27 hat Stefano Garzarella geschrieben:
> On Thu, Mar 28, 2019 at 03:42:49PM +0100, Kevin Wolf wrote:
> > Am 28.03.2019 um 14:35 hat Stefano Garzarella geschrieben:
> > > On Thu, Mar 28, 2019 at 12:59:55PM +0100, Kevin Wolf wrote:
> > > > Am 28.03.2019 um 11:52 hat Stefano Garzarella geschrieben:
> > > > > Several versions of GlusterFS (3.12? -> 6.0.1) fail when the
> > > > > transfer size is greater or equal to 1024 MiB, so we are
> > > > > limiting the transfer size to 512 MiB to avoid this rare issue.
> > > > >
> > > > > Buglink: https://bugzilla.redhat.com/show_bug.cgi?id=1691320
> > > > > Signed-off-by: Stefano Garzarella <address@hidden>
> > > > > ---
> > > > >
> > > > > RFC:
> > > > > Should I add a parameter to allow the user to modify the max_transfer
> > > > > variable?
> > > >
> > > > Ideally, gluster would have provided a way to detect automatically
> > > > whether the bug is present. But since you suggest a user option, I'm
> > > > afraid they haven't?
> > >
> > > Unfortunately, all version that I tried have this bug, also the latest
> > > available, for this reason, I hard coded the limit in the gluster driver.
> > > I also think that this limit is not reached very often for standard
> > > operations.
> > >
> > > Maybe I should check the version of the library during the configuration,
> > > when they will fix the issue.
> > >
> > > What do you suggest?
> >
> > Oh, sorry, I should have actually looked at the BZ. I assumed it was
> > already fixed, but it's not. Then I think what you have is fine for now.
>
> Good :)
> Should I resend without RFC or it can go as is through your block tree?
No need to resend, I'll apply it right away.
> > Maybe you could suggest to the gluster people that they add some kind of
> > a feature flag or something somewhere together with the fix so we can
> > find out at runtime whether the installed gluster libs are fixed or not.
>
> Thanks! I'll forward this useful suggestion!
Thanks!
Kevin