yahoo-eng-team team mailing list archive

[William <1625829@xxxxxxxxxxxxxxxxxx>]

Public bug reported:

Description:

After performing a shutdown of an instance and a start, the associated
floating IP is not reachable. A quick remedy is to delete a rule from an
associated security group and recreate the rule. Once the rule has been
toggled the instance is reachable again. Best guess is a race condition
in the neutron code. This error is non-deterministic.

This bug may be related to this bug -->
https://bugs.launchpad.net/neutron/+bug/1549443

Small test configuration:

A network with two attached instances on the same subnet is created. The
network is also connected to an openstack router with an interface to an
external network. Both instances on the network are assigned a floating
IP and are reachable via ssh and icmp. The instances are on separate
hypervisors. The bug may be isolated to the local instantiation of
neutron as it seems to only affect the instances on the same kvm node.

Steps to reproduce issue:

1. Shutdown one of the instances and then start it. 
2. Check for communication to the floating IP associated with that instance and the other instance.
3. No connectivity may be present. (If you can reach the instances, the race condition may not have been hit?)
4. Delete and recreate a rule on the default security group or any security group associated with the instance. 
5. Check for connectivity again.


Openstack version: Mitaka 8.1.2


In testing it seems that this output is not always the result. The attached file is the listing of iptables at corresponding steps in the process to reproduce the error.

** Affects: neutron
     Importance: Undecided
         Status: New

** Attachment added: "iptables-save from the kvm node"
   https://bugs.launchpad.net/bugs/1625829/+attachment/4744603/+files/BugReportiptables.txt

-- 
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to neutron.
https://bugs.launchpad.net/bugs/1625829

Title:
  instance shut down and start up breaks neutron networking

Status in neutron:
  New

Bug description:
  Description:

  After performing a shutdown of an instance and a start, the associated
  floating IP is not reachable. A quick remedy is to delete a rule from
  an associated security group and recreate the rule. Once the rule has
  been toggled the instance is reachable again. Best guess is a race
  condition in the neutron code. This error is non-deterministic.

  This bug may be related to this bug -->
  https://bugs.launchpad.net/neutron/+bug/1549443

  Small test configuration:

  A network with two attached instances on the same subnet is created.
  The network is also connected to an openstack router with an interface
  to an external network. Both instances on the network are assigned a
  floating IP and are reachable via ssh and icmp. The instances are on
  separate hypervisors. The bug may be isolated to the local
  instantiation of neutron as it seems to only affect the instances on
  the same kvm node.

  Steps to reproduce issue:

  1. Shutdown one of the instances and then start it. 
  2. Check for communication to the floating IP associated with that instance and the other instance.
  3. No connectivity may be present. (If you can reach the instances, the race condition may not have been hit?)
  4. Delete and recreate a rule on the default security group or any security group associated with the instance. 
  5. Check for connectivity again.

  
  Openstack version: Mitaka 8.1.2

  
  In testing it seems that this output is not always the result. The attached file is the listing of iptables at corresponding steps in the process to reproduce the error.

To manage notifications about this bug go to:
https://bugs.launchpad.net/neutron/+bug/1625829/+subscriptions