[ovs-dev] [Fix branch-2.4 travis builds] travis: Use container infrastructure.

[Andy Zhou]

On Thu, Mar 10, 2016 at 3:11 PM, Daniele Di Proietto <diproiettod at vmware.com
> wrote:

> Thanks for fixing this
>
> On 09/03/2016 19:55, "dev on behalf of Joe Stringer"
> <dev-bounces at openvswitch.org on behalf of joe at ovn.org> wrote:
>
> >Hi Andy, thanks for fixing this.
> >
> >On 9 March 2016 at 18:05, Andy Zhou <azhou at ovn.org> wrote:
> >> From: Daniele Di Proietto <diproiettod at vmware.com>
> >
> >As a general rule, it's nice to have the commit id direct in the
> >message to show the upstream commit being backported. I've been doing
> >it like this at the start of the message:
> >
> >Upstream: e88b97cbbec3497a2272ace805255d1c87230743
> >
> >This helps reviewers compare the original commit with the backport to
> >look for discrepancies.
> >
> >> Recently some testcases have been failing in travis because of a warning
> >> related to the use of an L3 device (OpenVZ specific) inside the workers.
>
> The fix is the same (moving to the container infrastructure), but the issue
> is different: the new "sudo" travis infrastructure has been moved to GCE,
> which
> doesn't have ipv6
>
Well, master builds still have some test with ipv6 and they are passing.

>
> https://blog.travis-ci.com/2015-11-27-moving-to-a-more-elastic-future
>
> Maybe you want to mention the problem in the backported commit message?
>
Sure, I will explain more in the back port section of the commit message.

>
>
> >>
> >> To get travis tests working again we can move to the newer container
> >> infrastructure: this commit does that.
> >>
> >> The disadvantage is that there's no sudo access anymore, but we can
> >> install packages with the apt plugin, and we shouldn't use root for
> >> anything else
> >>
> >> Also, since we're building DPDK with vhost-user (not vhost-cuse),
> >> libfuse-dev is not needed anymore.
>
> This was the case for master, but it's not the case for this backported
> commit. Do you think we should remove this line (I'm not sure what's the
> best procedure for backports).
>
I will keep the original message as is, since it is a back port.  In the
back port section of
comment, I'll add that for branch-2.4 builds, libfuse-dev is still needed.
Hope this is O.K.

>
> >>
> >> Tested-at: https://travis-ci.org/ddiproietto/ovs/builds/81764972
> >>
> >> [back ported to branch-2.4 by azhou]
> >>
> >> Signed-off-by: Daniele Di Proietto <diproiettod at vmware.com>
> >> CC: Joe Stringer <joestringer at nicira.com>
> >> Acked-by: Ben Pfaff <blp at nicira.com>
> >> Signed-off-by:  Andy Zhou <azhou at ovn.org>
> >
> ><snip>
> >
> >> @@ -26,4 +38,4 @@ script: ./.travis/build.sh $OPTS
> >>  notifications:
> >>    email:
> >>      recipients:
> >> -      - secure:
> >>KnZ6yDXDcC4VoiI04ZYR4sRTln7q16aXA7gVBa/M1jPWxl3BiTi+4idVE5bgrB1AK5iUwUXN6
> >>LQpjOdFDw1U/D+sKt+xmVG5MyLaTYIFp1TUOgtSGeiG3IUhpu125PN1i2EhXNqANyWyStCiIS
> >>DvJkDe4D/tbBehip1AEBuQONk=
> >> +      - build at openvswith.org
> >
> >This fragment wasn't in the original commit... was it intended?
> >(should it be a separate backport commit?)
>
> I think it should be OK to backport this on branch-2.4. Most of the
> travis infrastructure code should be kept in sync between the branches.
>
> It could be done in a separate commit, I'm happy either way
>
I will do this as a separate back port commit.

>
> Acked-by: Daniele Di Proietto <diproiettod at vmware.com>
>
>
> >
> >Other than this, LGTM.
> >
> >Acked-by: Joe Stringer <joe at ovn.org>
> >_______________________________________________
> >dev mailing list
> >dev at openvswitch.org
> >http://openvswitch.org/mailman/listinfo/dev
>
>