karaf logs.zip
odl1_log.tgz
odl2_log.tgz
odl3_log.tgz
[OVSDB-289] NetVirt Cluster Testing (odl-ovsdb-openstack) - Security Group Errors in karaf.log Created: 04/Feb/16 Updated: 03/May/18 Resolved: 09/May/16 |
|
| Status: | Resolved |
| Project: | ovsdb |
| Component/s: | openstack.net-virt |
| Affects Version/s: | unspecified |
| Fix Version/s: | None |
| Type: | Bug | ||
| Reporter: | Venkatrangan Govindarajan | Assignee: | ranjithkumar_t |
| Resolution: | Won't Do | Votes: | 0 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Environment: |
Operating System: All |
||
| Attachments: |
karaf logs.zip
odl1_log.tgz
odl2_log.tgz
odl3_log.tgz
|
| External issue ID: | 5223 |
| Description |
|
Performed simple operations using Openstack and netvirt as cluster Network Create and then created instances, the instances in Compute node not ping successfully. Errors in karaf.log a9-4c56-8299-64179f8d445d |
| Comments |
| Comment by Venkatrangan Govindarajan [ 04/Feb/16 ] |
|
Scenario ODL1 -> handle neutron requests ODL2/ODL3 -> Act as OVS Managers and OF Controllers to Openstack Initial test Both ODL2 and ODL3 were up -> the compute node instances were not able to reach dHCP Then ODL2 down, the compute node instances are working fine. Then ODL3 was made down and retried the cerrate networks + instances, the same phenomenon was observed again. |
| Comment by Venkatrangan Govindarajan [ 04/Feb/16 ] |
|
Observations During the first try RPC Exceptions seen in ODL2. During other trials 2016-02-03 20:01:53,719 | ERROR | ntDispatcherImpl | SecurityGroupCacheManagerImpl | 282 - org.opendaylight.ovsdb.openstack.net-virt - 1.2.1.SNAPSHOT | In processPortAdded cachedport port not found in neuton cache: cachedportUuid:01908c18-35e8-4e19-9923-2a3f3e86e2b6 exception was observed. |
| Comment by Venkatrangan Govindarajan [ 04/Feb/16 ] |
|
Attachment odl1_log.tgz has been added with description: ODL1 logss |
| Comment by Venkatrangan Govindarajan [ 04/Feb/16 ] |
|
Attachment odl2_log.tgz has been added with description: ODL2 logs |
| Comment by Venkatrangan Govindarajan [ 04/Feb/16 ] |
|
Attachment odl3_log.tgz has been added with description: ODL3 logs |
| Comment by Aswin Suryanarayanan [ 18/Feb/16 ] |
|
Venkat, Patch [1] solves issues with the restart/port delete scenarios. Could you please check whether the issue is solved with this patch? |
| Comment by Venkatrangan Govindarajan [ 18/Feb/16 ] |
|
Hi Aswin, I think while verifying for |
| Comment by Aswin Suryanarayanan [ 23/Feb/16 ] |
|
An NPE [1] is thrown on floating ip removing logic as mentioned in [2] when neutronL3Adapter.handleInterfaceEvent is called from SouthboundHandler which will prevent networkingProviderManager.getProvider(node).handleInterfaceDelete of SouthboundHandler being called. This should be causing an issue with clean up of SecurityGroupCache. The NPE is not caused by port being null. As there is sucessful reference of neutron port object before in the flow. So the floating ip list should have list of null entires. So fixing [2] should sovle this issue. [1] [2]https://bugs.opendaylight.org/show_bug.cgi?id=5290 |
| Comment by Aswin Suryanarayanan [ 24/Feb/16 ] |
|
The error log is observed in odl1 only Which was never shutdown. A node when restarted it will sync the security group cache with the current neutron port objects.So odl2 and odl3 does not have the error. |
| Comment by Aswin Suryanarayanan [ 21/Mar/16 ] |
|
Venkat, |
| Comment by ranjithkumar_t [ 09/May/16 ] |
|
Tested with stable/beryllium and master branch. There is no "Security Group" error in karaf log while making the ODL nodes down. It's working as expected. Hence closing this bug and attached the karaf logs. |
| Comment by ranjithkumar_t [ 09/May/16 ] |
|
Attachment karaf logs.zip has been added with description: karaf logs for all 3 ODL controllers |