Issue

• There are couple of dashboards created, that do some aggregations on fields in the logs. Some of these fields are not standard, eg kubernetes.labels.my-label.
• Elasticsearch keeps some meta data about fields in the documents of a certain type for each index. eg, field kubernetes.labels.my-label.
• These mappings seem to periodically reset/empty themselves, which breaks the dashboard.
• The fields kubernetes.pod_name, or kubernetes.node always make it back into the mapping which were disabled.
• The fields kubernetes.label.deploymentconfig, response.status, or request.headers.host all do not make it back in the mapping.