Another test that seems to fail often on jenkins but not other systems. We take down a jetty instance and then try to query a still up jetty index with the load balancing solr client (within the solrcloud client). We get an so read timeout on the request. I saw this once in early dev - I couldn't figure out what to blame other than http client - the other server was down, and somehow that was affecting the request to the server that was still up. I tried not sharing the httpclient instance between requests and making a new one each time and it started working - I reverted that and it was still working though - and worked since. Seems to fail consistently on jenkins though.