I managed to reproduce the issue with extended logs, and it turns out this happens when both hosts fail: the first one because of a timeout, the second one because of a fault. Now why it times out is the one million dollar question... See http://ci.hibernate.org/job/hibernate-search-yoann/org.hibernate$hibernate-search-elasticsearch/36/testReport/junit/org.hibernate.search.elasticsearch.test/DefaultElasticsearchClientFactoryTest/multipleHosts_failover_fault/ and the logs below.
|