The current nuxeo helm chart doesn't handle properly the cluster mode, for instance it is missing this configuration in nuxeo.conf:
We also need to enable the stickiness session on the Nginx used by the Ingress rule.
The current PVC policy (ReadWriteOnce) for logs or binaries doesn't allow several nuxeo to run at the same time.
Binaries could not be an issue as we'll probably leverage an external blob storage such as S3.
Logs should be individual for each Nuxeo, or we should enable the docker-json template and remove this PVC.