yahoo-eng-team team mailing list archive

[Mark McClain <1231930@xxxxxxxxxxxxxxxxxx>]

** No longer affects: neutron

-- 
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to neutron.
https://bugs.launchpad.net/bugs/1231930

Title:
  Rules dissapear after 300 seconds of inactivity

Status in tripleo - openstack on openstack:
  Fix Released

Bug description:
  Using Feodra 19 for TripleO

  In the TripleO overcloud I can successfully connect to VM's but soon
  after loose ssh connections,

  I think this is after at least 5 minutes of inactivity, I'm observing
  the learned rule in table 20 vanish if the hard_age hits 300, see the
  two snippets from "ovs-ofctl dump-flows br-tun" with the learned rule
  and then gone the next second

  
  Fri 27 Sep 09:49:14 UTC 2013

   cookie=0x0, duration=1930.786s, table=10, n_packets=247, n_bytes=32673, idle_age=300, priority=1 actions=learn(table=20,hard_timeout=300,priority=1,NXM_OF_VLAN_TCI[0..11],NXM_OF_ETH_DST[]=NXM_OF_ETH_SRC[],load:
  0->NXM_OF_VLAN_TCI[],load:NXM_NX_TUN_ID[]->NXM_NX_TUN_ID[],output:NXM_OF_IN_PORT[]),output:1
   cookie=0x0, duration=800.717s, table=20, n_packets=231, n_bytes=23336, hard_timeout=300, idle_age=300, hard_age=300, priority=1,vlan_tci=0x0002/0x0fff,dl_dst=fa:16:3e:e4:de:d6 actions=load:0->NXM_OF_VLAN_TCI[],load:0x1->NXM_NX_TUN_ID[],output:2
   

  
  Fri 27 Sep 09:49:15 UTC 2013

   cookie=0x0, duration=1931.798s, table=10, n_packets=247,
  n_bytes=32673, idle_age=301, priority=1
  actions=learn(table=20,hard_timeout=300,priority=1,NXM_OF_VLAN_TCI[0..11],NXM_OF_ETH_DST[]=NXM_OF_ETH_SRC[],load:0->NXM_OF_VLAN_TCI[],load:NXM_NX_TUN_ID[]->NXM_NX_TUN_ID[],output:NXM_OF_IN_PORT[]),output:1

  Is the correct thing to do just to remove the hard_timeout=300, this
  seems to work for me

  diff --git a/neutron/plugins/openvswitch/agent/ovs_neutron_agent.py b/neutron/plugins/openvswitch/agent/ovs_neutron_agent.py
  index eefe384..62c87e3 100644
  --- a/neutron/plugins/openvswitch/agent/ovs_neutron_agent.py
  +++ b/neutron/plugins/openvswitch/agent/ovs_neutron_agent.py
  @@ -715,7 +715,6 @@ class OVSNeutronAgent(sg_rpc.SecurityGroupAgentRpcCallbackMixin,
           # adresses (assumes that lvid has already been set by a previous flow)
           learned_flow = ("table=%s,"
                           "priority=1,"
  -                        "hard_timeout=300,"
                           "NXM_OF_VLAN_TCI[0..11],"
                           "NXM_OF_ETH_DST[]=NXM_OF_ETH_SRC[],"
                           "load:0->NXM_OF_VLAN_TCI[],"

  
  or should this rule just reappear when I try to reconnect, I've also observed the rule returning when I try to connect, and I can then connect so the loss of connectivity doesn't always happen after 5 minutes. But the rule not reappearing seems to line up with times I can't ssh to the VM.

  Also its worth noting that the hard_timeout is whats being set but it
  appears to be acting more like a idle_timeout although I may be
  understanding something wroge here, ovs newbie...

To manage notifications about this bug go to:
https://bugs.launchpad.net/tripleo/+bug/1231930/+subscriptions