[ https://issues.apache.org/jira/browse/DRILL-4433?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15163528#comment-15163528 ]
Jacques Nadeau commented on DRILL-4433: --------------------------------------- I would add an additional note here, as well: ZK is only used by the client to determine what nodes are available, not maintain cluster cohesion. As such, it should be fast and not a long-lasting connection. I think that as part of this change we should also use a faster mechanism to doing this lookup. > Unnecessary entries created in ZK when a jdbc application uses a wrong > connection URL > -------------------------------------------------------------------------------------- > > Key: DRILL-4433 > URL: https://issues.apache.org/jira/browse/DRILL-4433 > Project: Apache Drill > Issue Type: Bug > Components: Client - JDBC > Reporter: Rahul Challapalli > > commit # : 6d5f4983003b8f5d351adcb0bc9881d2dc4c2d3f > commit date : Feb 22, 2016 > In my JDBC application, I accidentally used an invalid connection string like > below > {code} > jdbc:drill:zk=x.x.x.x:5181/zkroot-blah/cluster-name > {code} > While drill correctly reported an "No DrillbitEndpoint can be found error", I > also observed that it created an entry in zookeeper by the name > "zkroot-blah". There seems to be no reason for doing this. -- This message was sent by Atlassian JIRA (v6.3.4#6332)