You are viewing a plain text version of this content. The canonical link for it is here.

Posted to user@phoenix.apache.org by Kumar Palaniappan <kp...@marinsoftware.com> on 2017/09/28 18:37:08 UTC

Connection issue with Spark/Phoenix/ZK

After upgrading to CDH 5.9.1/Phoenix 4.10/Spark 1.6 from CDH 5.5.2/Phoenix
4.6/Spark 1.5, streaming jobs that read data from Phoenix no longer release
their zookeeper connections, meaning that the number of connections from
the driver grow with each batch until the ZooKeeper limit on connections
per IP address is reached, at which point the Spark streaming job can no
longer read data from Phoenix.

Appreciate the pointers.