yahoo-eng-team team mailing list archive

[KaiLin <linkai3@xxxxxxxxxx>]

Public bug reported:

ENV:
3 controllers
Juno OpenStack

SYMPTOM:
1.create a vm on a vxlan network
2.login the vm, cannot contact with it
3.check flows in br-tun,the vm's income flows in br-tun is lost

CAUSE:
1.In neutron-openvswitch-agent log, the error info is: 
in _del_fdb_flow lvm.tun_ofports.remove(ofport) KeyError:'28'
the reason is: the ofport=28 flow is deleted twice.The following will say why delete twice.
2.The error will cause the lock flag of flow is open, and the delete operator is placed in the add process.so after create a vm, the normal flow table just issued, but is deleted by before abnormal flow operating.

why the flow is deleted twice? When delete VM, nova delete the tap device, and the ovs-agent scan the change, then do rpc handling(l2pop fdb_remove) in server1.
Then nova will send delete port to neutron, neutron server will do rpc handling again in server2.

** Affects: neutron
     Importance: Undecided
     Assignee: KaiLin (linkai3)
         Status: New

** Changed in: neutron
     Assignee: (unassigned) => KaiLin (linkai3)

** Summary changed:

- Delete VM low probablity lead to vxlan upstream flow lost
+ Delete VM low probablity lead to vxlan flow lost

-- 
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to neutron.
https://bugs.launchpad.net/bugs/1430127

Title:
  Delete VM low probablity lead to vxlan flow lost

Status in OpenStack Neutron (virtual network service):
  New

Bug description:
  ENV:
  3 controllers
  Juno OpenStack

  SYMPTOM:
  1.create a vm on a vxlan network
  2.login the vm, cannot contact with it
  3.check flows in br-tun,the vm's income flows in br-tun is lost

  CAUSE:
  1.In neutron-openvswitch-agent log, the error info is: 
  in _del_fdb_flow lvm.tun_ofports.remove(ofport) KeyError:'28'
  the reason is: the ofport=28 flow is deleted twice.The following will say why delete twice.
  2.The error will cause the lock flag of flow is open, and the delete operator is placed in the add process.so after create a vm, the normal flow table just issued, but is deleted by before abnormal flow operating.

  why the flow is deleted twice? When delete VM, nova delete the tap device, and the ovs-agent scan the change, then do rpc handling(l2pop fdb_remove) in server1.
  Then nova will send delete port to neutron, neutron server will do rpc handling again in server2.

To manage notifications about this bug go to:
https://bugs.launchpad.net/neutron/+bug/1430127/+subscriptions