Uploaded image for project: 'OpenShift SDN'
  1. OpenShift SDN
  2. SDN-5505

kubevirt live migration with primary UDN break if VMs are the clients using node ip as src ip.

XMLWordPrintable

    • Icon: Story Story
    • Resolution: Unresolved
    • Icon: Undefined Undefined
    • None
    • None
    • None
    • Strategic Product Work
    • False
    • None
    • False
    • OCPSTRAT-1613 - Enhance User Defined Networks: Add support for Services(Full), EIP(Full), NetPol: GA
    • ---
    • 0
    • 0
    • Rejected

      Description of problem:

      When a kubevirt vm is using primary udn layer2 network and opening a connection towards and external service, after live migration the traffic is going to egress over different node so src-ip changes breaking the 5 tuple tcp connections and making server send a RST since src-ip has changed.

      Version-Release number of selected component (if applicable): 4.18

      How reproducible: Always

      Steps to Reproduce:

      1. Create an external container with something like "iperf3 -s -B external ip"

      2. Create a VM with layer2 primary UDN running "iperf3 -c external ip -t 0"

      3. Live migrate the VM

      Actual results:

      The client receive a TCP RST and iperf3 client breaks with "peer reset"

      Expected results:

      Tcp connection is not broken

      Additional info:

      At other virt environments like openstack this is done using the "floating ip" feature from OVN perspective they use "dnat_and_snat" but that do not work currently with interconnect, it may work with future transit router work.

       

      Affected Platforms: internal CI failuter

              bbennett@redhat.com Ben Bennett
              ellorent Felix Enrique Llorente Pastora
              Anurag Saxena Anurag Saxena
              Votes:
              0 Vote for this issue
              Watchers:
              3 Start watching this issue

                Created:
                Updated: