Elasticsearch: exception caught on transport layer
Make sure to keep the elasticsearch client library jar in sync with the version of your cluster.
For example when using Maven:
<dependency> <groupId>org.elasticsearch</groupId> <artifactId>elasticsearch</artifactId> <version>2.1.0</version> </dependency>
This happens for any >2.x Elasticsearch instance anytime there is still a 1.x machine looking for clusters (it sends bad protocol information) on the same network. Technically you can ignore it (it's just letting you know that the 1.x machine tried to connect but did an invalid connection), however probably your logs will fill up etc. if you leave it alone too long.
Best solution is to isolate your old 1.x nodes from the newer nodes until you're done upgrading.