In the case of docker swarm cluster, the rabbitmq starts on all the nodes as global. This will cause portal to write MQ data to all the nodes which can cause inconsistency.
For that reason we need to scale down the rabbitmq to run on single node.
All Supported Portal versions.
a. run : `portal.sh keep`
b. Above will leave docker.compose file in place.
c. Edit the file and update the rabbitmq definition by removing "mode: global" and replacing with replicas: 1
Remove
rabbitmq:
deploy:
mode: global
Replace
rabbitmq:
deploy:
replicas: 1
d. Stop portal: docker stack rm portal
e. Start portal manually: docker stack deploy --with-registry-auth --compose-file docker-compose.yml --compose-file analytics.yml portal
f. Run 'docker service ls' and check portal_rabbitmq mode is now replicated and Relicas are 1/1.
g. If the replicas are 2/1, then run this command to update the changes.
# docker service update --force <rabbitmq_ID>