Note: This bug is displayed in read-only format because the product is no longer active in Red Hat Bugzilla.

Bug 1663265

Summary: [OVN] connectivity between floating ips stopped after rebooting controller nodes
Product: Red Hat OpenStack Reporter: Roman Safronov <rsafrono>
Component: python-networking-ovnAssignee: Assaf Muller <amuller>
Status: CLOSED DUPLICATE QA Contact: Eran Kuris <ekuris>
Severity: medium Docs Contact:
Priority: unspecified    
Version: 14.0 (Rocky)CC: apevec, lhh, lmartins, majopela, nyechiel
Target Milestone: ---   
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: If docs needed, set a value
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2019-01-15 13:31:07 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:

Description Roman Safronov 2019-01-03 15:27:43 UTC
Description of problem:
No ping between 2 instances of 2 different tenants via floating IPs after rebooting controllers.

Version-Release number of selected component (if applicable):
Red Hat OpenStack Platform release 14.0.0 RC (Rocky)
core_puddle=2018-12-20.4

openvswitch2.10-ovn-common-2.10.0-28.el7fdp.1.x86_64
openvswitch2.10-ovn-central-2.10.0-28.el7fdp.1.x86_64
rhosp-openvswitch-ovn-common-2.10-0.1.el7ost.noarch
rhosp-openvswitch-ovn-central-2.10-0.1.el7ost.noarch


How reproducible:
Ocassionally. I tried to reproduce once again without success.

Steps to Reproduce:

1. Prepare OSP14 HA setup with 3 controller and 2 compute nodes. Enable OVN and DVR. 

2.Set 'allow_overlapping_ips=True' in /etc/neutron/neutron.conf on all controller nodes (inside neutron_api containers) and restart the containers.

3.Create 2 tenants. For each tenant create a network, a subnet (in my case it was overlapping subnet, i.e. 192.168.3.0/24 for both), router between the tenant internal network and public network. Run an instance in each tenant and create floating IPs for them. Allow ssh and icmp for used security group.

4. From both of the instances try to ping floating IP address of the other instance.
-->Ping echo replies received - OK

4. Restart all controller nodes simultaneously. Wait until all controllers are up and running.

5. Shutdown master controller.[Note: I used 'virsh shutdown' command]. Then start it back.

6. From both of the instances try to ping floating IP address of the other instance.


Actual results:
No ping replies from other instance floating IP - BUG 

Expected results:
There are ping replies from floating IP of an instance on other compute node. 

Additional info:
Numan already investigated the issue. After destroying some mac_bindings in Southbound DB ping between floating IPs works again.

The issue was found during executing scenario from test: RHELOSP-26294 - controller - recovery
from OVN Neutron HA test plan:
https://polarion.engineering.redhat.com/polarion/#/project/RHELOpenStackPlatform/wiki/OVN/OVN%20Neutron%20-%20HA

Comment 1 Roman Safronov 2019-01-03 16:29:37 UTC
The sosreports available here: http://rhos-release.virt.bos.redhat.com/log/bz1663265

Comment 2 Lucas Alvares Gomes 2019-01-15 13:31:07 UTC

*** This bug has been marked as a duplicate of bug 1643900 ***