Kevin Ludwig created SOLR-4211: ---------------------------------- Summary: LBHttpSolrServer Key: SOLR-4211 URL: https://issues.apache.org/jira/browse/SOLR-4211 Project: Solr Issue Type: New Feature Components: clients - java Affects Versions: 4.0 Reporter: Kevin Ludwig Priority: Minor
I would like for SOLRJ's LBHttpSolrServer to support graceful shutdown of SOLR machines. SOLR's PingRequestHandler (e.g. "/admin/ping") already has support for healthcheck files, and LBHttpSolrServer already has a ping() method that calls this endpoint. Recent changes in LBHttpSolrServer introduced the notion of an alive list and a zombie list, as well as a background thread to check for dead nodes that are back alive. My proposal is to have the background thread: 1. determine if nodes are alive via ping() rather than query("*:*"). 2. also check for alive servers that have gone out of service (again, via ping()). Also the basic logic in the public request method is to try all alive nodes, and if none are reachable then try each zombie. If a node is brought offline (via removing healthcheck file, causing ping() to fail) then this retry should not be done. I'm willing to submit a patch for this if needed. -- This message is automatically generated by JIRA. If you think it was sent incorrectly, please contact your JIRA administrators For more information on JIRA, see: http://www.atlassian.com/software/jira --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org