Missing milli/nano-seconds in logs shipped to elasticsearch

Hi all, I have enabled project logging in Rancher and the container logs are forwarded by rancher (using fluentd) to an Elasticsearch and Kibana environment. When I look at the logs in Kibana, all the log messages only have second accuracy. This means the log messages are all out of order within one second and makes it unusable for debugging purposes. I have checked the docker logs and the contain nanosecond resolution. It seems that the rancher/fluentd configuration skips the nanosecond part from the timestamp. A little more investigation in the configuration (project.log) of fluentd which is created by Rancher I can see that the “time_format” which is used has value “%Y-%m-%dT%H:%M:%S”. How can I configure rancher or change the fluentd template that it uses the full ISO8601 date/time format including nanoseconds to have the exact logtime in elasticsearch.