Re: [Xen-devel] [xen-unstable test] 100789: regressions - FAIL [and 2 more messages]

[Ian Jackson <ian.jackson@xxxxxxxxxxxxx>]

Wei Liu writes ("Re: [Xen-devel] [xen-unstable test] 100789: regressions - 
FAIL"):
> I see three ways to move this forward.
>
> 3. Retire these two tests.

Do we expect users still to want VHD support ?  We still allegedly
support VHD for guests.  So we shouldn't retire these tests unless we
are dropping VHD support entirely.

AFAICT users who want VHD support need to be able to create images
etc.

> 2. Install blktap-utils shipped in Debian (available from Wheezy
>    onwards), the main difficulty would be the package depends on a dkms
>    package that seems to require building with kernel header when
>    installing.

According to Debian, we (Xen) are upstream for this package.  It makes
no sense for osstest to install something from Debian which we have
deleted upstream !

> 1. Resurrect vhd-util from blktap2.

What is wrong with this plan ?

> In the meantime, if we want to avoid blocking xen-unstable for too long,
> we might want to force push.

If that's a consideration, we should be considering a revert, not a
force push.

Andrew Cooper writes ("Re: [Xen-devel] [xen-unstable test] 100789: regressions 
- FAIL"):
> +1 to a force push for now.  There are quite a few changes currently
> blocked.

IMO that is not a good reason for a force push.

Wei Liu writes ("Re: [Xen-devel] [xen-unstable test] 100789: regressions - 
FAIL"):
> 4. Provide a pre-made vhd image.
> 
> vhd-util create disk.vhd -s 10000 -> 24K in actual size.

This is no good because

1. disk.vhd is a file for which we would have deleted the source code!

2. Users need the ability to create images.

Ian.

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel