bagpipe-bgp : MPLS label allocator should not recycle too fast

[Thomas Morin]

Reusing an MPLS label soon after a VPN instance stopped using it, opens the door to the case where another VPN instace (on a local or remote BGP speaker) may still be using it for its previous semantic because it hasn't processed the BGP events.

(issue present since [1], not present in a released version)

[1] https://review.openstack.org/#/c/561946/

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/563665
Committed: https://git.openstack.org/cgit/openstack/networking-bagpipe/commit/?id=7a2f332e1b49f9418b913598f4162e5828f4da6a
Submitter: Zuul
Branch: master

commit 7a2f332e1b49f9418b913598f4162e5828f4da6a
Author: Thomas Morin <email address hidden>
Date: Mon Apr 23 17:15:16 2018 +0200

    bagpipe-bgp: ID allocator to reuse an id as late as possible

    With this change the label allocator will not reuse
    an ID if possible to find one subsequent to the last
    allocated.

    Additionally, to catch potential inconsistencies,
    releasing a non-allocated id will raise an exception,
    rather than just log a warning.

    Change-Id: Iba1f6573985a0edaff2f91909ac257a9bc464b7b
    Closes-Bug: 1766239

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/networking-bagpipe 9.0.0.0b2 development milestone.