periodic-tripleo-ci-centos-7-multinode-1ctlr-featureset030-master vxlan tunnel fails randomly

[Gabriele Cerami]

logs at

https://logs.rdoproject.org/openstack-periodic/periodic-tripleo-ci-centos-7-multinode-1ctlr-featureset030-master/a120df1/console.txt.gz#_2018-04-26_06_40_50_208

show that ping across nodes fails even after restarting ovs.

This is an intermittent issue, but we have various reports on the same issue from both jobs and user that tried to reproduce multinode on rdocloud.

[Gabriele Cerami (gcerami)]

The tunnel does not work because the VXLAN encapsulation packet on port 4789 are blocked by the security groups associated with the nodes. Once we opened the UDP port, the ping worked correctly

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/564475

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/564475
Committed: https://git.openstack.org/cgit/openstack/tripleo-quickstart-extras/commit/?id=e06f50fee7e5b551ebc3ed902b4ee37eadcf6ec1
Submitter: Zuul
Branch: master

commit e06f50fee7e5b551ebc3ed902b4ee37eadcf6ec1
Author: Enrique Llorente <email address hidden>
Date: Thu Apr 26 14:08:55 2018 +0200

    Add vxlan port to the multinodes security group

    There are problems with RDO cloud + multinode communication, there is no
    vxlan package arriving from subnode-1 to subnode-0, to allow that vxlan
    port needs to be open.

    Partial-Bug: 1767099
    Change-Id: I9d070e208000194e73654299c9645679bea27169

[Sagi (Sergey) Shnaidman (sshnaidm)]

@Gabriele, was port opening in cloud documented or configured anywhere? How do we know to avoid this next time?

[Jose Luis Franco (jfrancoa)]

@Gabriele, can we consider this bug as closed? I didn't see this error happening anymore after Quique's patch got merged. (The lp is still opened as the patch had the "Partial-Bug" tag, not "Closes-Bug")

[Alex Schultz (alex-schultz)]

Closing this out, feel free to reopen if it comes back up as an issue.