[MRE] Iscsitarget 1.4.20.3+svn490 into Precise

[Stefan Bader]

On 20.12.2013 22:15, Steve Langasek wrote:
> Hi Stefan,
> 
> On Thu, Dec 19, 2013 at 04:31:39PM +0100, Stefan Bader wrote:
>> https://bugs.launchpad.net/ubuntu/+source/iscsitarget/+bug/1262712
> 
>> While looking into the required kernel parts it seemed to be a reasonable move
>> to actually backport the complete source package as those seemed to be bugfixes
>> to the tools only.
> 
>> Normally this would be presented to the technical-board if that was still (or
>> again) present. This is unfortunately a bit last minute but should go into
>> Precise before finalizing the 12.04.4.
> 
> In the absence of a TB, we can still discuss how this package fits the MRE
> rules.
> 
> Generally, MREs are granted for packages that have upstream bugfix-only
> branches, that undergo upstream QA (or include thorough test suites) so that
> we can feel confident that the update won't cause regressions, without
> having to subject the changes to the kind of close scrutiny that we give to
> other SRUs.
> 
> It sounds like this iscsitarget branch may meet the requirement of a
> bugfix-only branch; can you confirm that the 1.4.20.x series is *guaranteed*
> to be bugfix-only, or is that just current practice?  If it's not
> guaranteed, we might prefer to make a one-time exception instead of having a
> standing micro-release exception.

James Page was dealing with upstream a bit closer. But one thing first, this was
actually meant as a one-time MRE. Not a general one. So the whole decision is
based on whether this exact update looks safe. And I had been comparing the
changes in the userspace code and found nothing that looked like a new feature.
The kernel-side also (iirc) was only changes to keep up with upstream kernel
changes that cause the compile to break. And I believe they were also arranged
in a manner that made changes only when being compiled to a newer kernel version
that requires a fix.

> 
> The other point is QA.  If we aren't planning to individually test the
> upstream changes for regressions, how do we verify the backport as a whole? 
> The testcase you include in the bug looks to me like a good smoke test, but
> I would be worried that it's not broad enough to catch all probable issues.

I may be a bit prejudiced here with my technical hat on. The main reason to get
the package backported to precise was, that since this version was released with
Saucy it has at least some broader testing in real world. The kernel side has
not changed much (or not at all feature wise). So I would expect issues rather
while interacting with user-space. Which I tried to cover. Maybe there is
something which can be done simple and quick (Ok, since is a dkms package it
probably probably does not need to be on the images and so has a bit more time)
but as soon as the 12.04.4 images go out and people use the Saucy kernels in
Precise we would have a problem. James, which I copied, may know whether the
server team has some QA which involves the iscsitarget in some way.

-Stefan
> 
> Thanks,
> 


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 901 bytes
Desc: OpenPGP digital signature
URL: <https://lists.ubuntu.com/archives/ubuntu-release/attachments/20131223/6bb5d869/attachment.pgp>