Interface drivers don't update port MTU if the port already exists

[Ihar Hrachyshka]

This is needed because Neutron allows to change MTU values for networks (through configuration options modification and neutron-server restart). Without that, there is no way to apply new MTU for DHCP and router ports without migrating resources to other nodes.

I suggest we apply MTU on conseqent plug() attempts, even if port exists.

[Ihar Hrachyshka (ihar-hrachyshka)]

Neutron fix: https://review.openstack.org/405532

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/405532
Committed: https://git.openstack.org/cgit/openstack/neutron/commit/?id=5c8dffa7fb6c95a04a7b50c7d6e63c9a2729231b
Submitter: Jenkins
Branch: master

commit 5c8dffa7fb6c95a04a7b50c7d6e63c9a2729231b
Author: Ihar Hrachyshka <email address hidden>
Date: Tue Nov 29 22:24:29 2016 +0000

    Update MTU on existing devices

    This patch makes OVS and Linuxbridge interface drivers to set MTU on
    plug() attempt if the device already exists. This helps when network MTU
    changes (which happens after some configuration file changes).

    This will allow to update MTU values on agent restart, without the need
    to bind all ports to new nodes, that would involve migrating agents. It
    will also help in case when you have no other nodes to migrate to (in
    single node mode).

    Both OVS and Linuxbridge interface drivers are updated.

    Other drivers (in-tree IVS as well as 3party drivers) will use the
    default set_mtu implementation, that only warns about the missing
    feature (once per process startup).

    DocImpact suggest to restart agents after MTU config changes instead of
              rewiring router/DHCP ports.

    Related: If438e4816b425e6c5021a55567dcaaa77d1fffff
    Related: If09eda334cddc74910dda7a4fb498b7987714be3
    Closes-Bug: #1649845
    Change-Id: I3c6d6cb55c5808facec38f87114c2ddf548f05f1

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: stable/newton
Review: https://review.openstack.org/412462

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/os-vif 1.4.0 release.

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/412462
Committed: https://git.openstack.org/cgit/openstack/neutron/commit/?id=52f31b0ddbb978402f65850bf41ceb409c560d4f
Submitter: Jenkins
Branch: stable/newton

commit 52f31b0ddbb978402f65850bf41ceb409c560d4f
Author: Ihar Hrachyshka <email address hidden>
Date: Tue Nov 29 22:24:29 2016 +0000

    Update MTU on existing devices

    This patch makes OVS and Linuxbridge interface drivers to set MTU on
    plug() attempt if the device already exists. This helps when network MTU
    changes (which happens after some configuration file changes).

    This will allow to update MTU values on agent restart, without the need
    to bind all ports to new nodes, that would involve migrating agents. It
    will also help in case when you have no other nodes to migrate to (in
    single node mode).

    Both OVS and Linuxbridge interface drivers are updated.

    Other drivers (in-tree IVS as well as 3party drivers) will use the
    default set_mtu implementation, that only warns about the missing
    feature (once per process startup).

    DocImpact suggest to restart agents after MTU config changes instead of
              rewiring router/DHCP ports.

    Related: If438e4816b425e6c5021a55567dcaaa77d1fffff
    Related: If09eda334cddc74910dda7a4fb498b7987714be3
    Closes-Bug: #1649845
    Change-Id: I3c6d6cb55c5808facec38f87114c2ddf548f05f1
    (cherry picked from commit 5c8dffa7fb6c95a04a7b50c7d6e63c9a2729231b)

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/neutron 10.0.0.0b3 development milestone.

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/neutron 9.2.0 release.

[Sean Dague (sdague)]

There are no currently open reviews on this bug, changing the status back to the previous state and unassigning. If there are active reviews related to this bug, please include links in comments.

[Sean Dague (sdague)]

Found open reviews for this bug in gerrit, setting to In Progress.

review: https://review.openstack.org/408781 in branch: master

[OpenStack Infra (hudson-openstack)]

Change abandoned by Stephen Finucane (<email address hidden>) on branch: master
Review: https://review.openstack.org/408781
Reason: Turns out plugging hybrid OVS VIFs was moved to os-vif in commit '745f5fbb', so this does nothing. This can be seen in the failing tests.

I've pushed the version with tests in case anyone wants the backport for Mitaka and earlier on their own branch. However, Mitaka is no longer support upstream so we can abandon this

[Sean Dague (sdague)]

There are no currently open reviews on this bug, changing the status back to the previous state and unassigning. If there are active reviews related to this bug, please include links in comments.

[Sean Dague (sdague)]

This is apparently fixed in os-vif for Newton and beyond. Marking Invalid on the Nova side because the logic doesn't live in Nova in any supported version.