alerman opened a new issue, #4303: URL: https://github.com/apache/accumulo/issues/4303
**Describe the bug** When monitoring zookeeper connections on a large cluster the shell makes 2 * ( number of tservers) + 1 connections to zookeeper on connect when trying to determine tablet server to connect to. It looks like it is retrieving from ZK: 1. The node that contains all tservers ( in our case `/accumulo/<instance id>/tservers`) 2. The list of locks under `/accumulo/<instance id>/tservers/<server id>` 3. The lock at `/accumulo/<instance id>/tservers/<server id>/<lock id>` It would make sense to cut this off once it has a valid tserver, until such time as talking to other tservers is necessary. **Versions (OS, Maven, Java, and others, as appropriate):** Affected Version - 2.1.2 **To Reproduce** 1. Enable ZK logging at DEBUG 2. Connect to shell, monitoring ClientCnxn messages from zk 3. Observe requests to ZK for each tserver **Expected behavior** Accumulo should only make as many connections as necessary to function. Making extra calls wastes time and resources. **Additional context** This problem would probably not have a noticable affect on small systems. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
