Sometimes when contrail is unable to push configuration changes to its managed Fabric devices it is worth to check the config-api and rabbitmq status.

• contrail is unable to push configuration changes to its managed Fabric devices.



• Seeing these error messages in /var/log/contrail/config-api/contrail-api-0.loglogs

[contrail-api] [WARNING]: __default__ [SYS_WARN]: VncApiError: KombuAmqpClient: Connection error in Kombu amqp publisher greenlet: Basic.return: (312) NO_ROUTE 

• huge number of connection attempts to rabbitmq from config-api workers - on controller verify the rabbitmq connections using "netstat -antp | grep 5673 | wc -l" (ideally this value will be less than 100).

• on any one controller, check rabbbitmq status using "rabbitmqctl cluster_status" and "rabbitmqctl list_queues" (In rabbitmqctl cluster_status we should see all the controller nodes under "Running Nodes" and under "Maintenance status" all controller nodes should show "not under maintenance". Also, in "rabbitmqctl list_queues" we should not see huge number of "devicemgr" queues)

• on controllers, we should not see these error messages (that confirms rabbitmq and config-api-workers are successfully able to connect) here - /var/log/contrail/config-api/contrail-api-0.log.

"[contrail-api] [WARNING]: __default__ [SYS_WARN]: VncApiError: KombuAmqpClient: Connection error in Kombu amqp publisher greenlet: Basic.return: (312) NO_ROUTE"

If these symptoms are seen solution is to restart config-api and rabbitmq services on contrail controllers.