[Bug 1964348] [NEW] [20.03][20.12] ovn-controller should update OF rules atomically

[Frode Nordahl]

Public bug reported:

Originally reported at [0] and fixed in OVN 21.06 by [1].

If some OF rules needs to be replaced due to logical flow changes,
ovn-controller deletes old rules first and adds new ones later.
In a complex scenario with big number of flows a lot of time
can pass between these events leading to the dataplane downtime
and packet loss.  Also, while these changes are in progress,
OVS will use incomplete pipelines that will also lead to packet
drops.

To avoid this, all flow modifications should be done atomically,
so there will be always some version of OF rules installed that
can handle dataplane traffic and it will be complete in terms of
reflecting some consistent set of logical flows.  Wrapping all
flow modifications into atomic ordered bundle to achieve that.

0: https://bugzilla.redhat.com/1947398
1: https://github.com/ovn-org/ovn/commit/7687d878f1e30c98d52b37bade42293f2a1e9bd3

** Affects: cloud-archive
     Importance: Undecided
         Status: New

** Affects: cloud-archive/wallaby
     Importance: Undecided
         Status: New

** Affects: ovn (Ubuntu)
     Importance: Undecided
         Status: Fix Released

** Affects: ovn (Ubuntu Focal)
     Importance: Undecided
         Status: New

** Also affects: ovn (Ubuntu Focal)
   Importance: Undecided
       Status: New

** Changed in: ovn (Ubuntu)
       Status: New => Fix Released

** Also affects: cloud-archive
   Importance: Undecided
       Status: New

** Also affects: cloud-archive/wallaby
   Importance: Undecided
       Status: New

-- 
You received this bug notification because you are a member of Ubuntu
OpenStack, which is subscribed to ovn in Ubuntu.
https://bugs.launchpad.net/bugs/1964348

Title:
  [20.03][20.12] ovn-controller should update OF rules atomically

Status in Ubuntu Cloud Archive:
  New
Status in Ubuntu Cloud Archive wallaby series:
  New
Status in ovn package in Ubuntu:
  Fix Released
Status in ovn source package in Focal:
  New

Bug description:
  Originally reported at [0] and fixed in OVN 21.06 by [1].

  If some OF rules needs to be replaced due to logical flow changes,
  ovn-controller deletes old rules first and adds new ones later.
  In a complex scenario with big number of flows a lot of time
  can pass between these events leading to the dataplane downtime
  and packet loss.  Also, while these changes are in progress,
  OVS will use incomplete pipelines that will also lead to packet
  drops.

  To avoid this, all flow modifications should be done atomically,
  so there will be always some version of OF rules installed that
  can handle dataplane traffic and it will be complete in terms of
  reflecting some consistent set of logical flows.  Wrapping all
  flow modifications into atomic ordered bundle to achieve that.

  0: https://bugzilla.redhat.com/1947398
  1: https://github.com/ovn-org/ovn/commit/7687d878f1e30c98d52b37bade42293f2a1e9bd3

To manage notifications about this bug go to:
https://bugs.launchpad.net/cloud-archive/+bug/1964348/+subscriptions