Nitul created DRILL-8427:
----------------------------
Summary: Apache drill connectivity with zookeeper on Kubernetes
cluster with JDBC data source with Kubernetes or docker
Key: DRILL-8427
URL: https://issues.apache.org/jira/browse/DRILL-8427
Project: Apache Drill
Issue Type: Bug
Components: Client - JDBC
Affects Versions: 1.17.0
Reporter: Nitul
Setup:
Java application can use specified JDBC driver to connect to drill via
zookeeper. https://drill.apache.org/docs/using-the-jdbc-driver/
E.g. jdbc:drill:zk=10.10.100.56:2181/drill/drillbits1;schema=hive
Using this zookeeper provides a host name of the drill which is available and
ready to serve the query.
Java app will connect to this drill host and gets the connection to execute the
query.
In case of Kubernetes drill pod is running and registered to zookeeper with
some dynamic pod name. Something like apachedrill-cb97cfc7c-mfkhr.
Now, zookeeper provides apache drill's hostname as above and java app is trying
to connect to this pod, but this pod host name is resolved and it failed.
{code:java}
java.sql.SQLNonTransientConnectionException: Failure in connecting to Drill:
oadd.org.apache.drill.exec.rpc.RpcException: CONNECTION :
java.net.UnknownHostException: apachedrill-cb97cfc7c-mfkhr
\tat
org.apache.drill.jdbc.impl.DrillConnectionImpl.<init>(DrillConnectionImpl.java:164)
\tat
org.apache.drill.jdbc.impl.DrillJdbc41Factory.newDrillConnection(DrillJdbc41Factory.java:67)
\tat org.apache.drill.jdbc.impl.DrillFactory.newConnection(DrillFactory.java:67)
\tat
oadd.org.apache.calcite.avatica.UnregisteredDriver.connect(UnregisteredDriver.java:138)
\tat org.apache.drill.jdbc.Driver.connect(Driver.java:75)
\tat
com.zaxxer.hikari.util.DriverDataSource.getConnection(DriverDataSource.java:138)
\tat com.zaxxer.hikari.pool.PoolBase.newConnection(PoolBase.java:364)
\tat com.zaxxer.hikari.pool.PoolBase.newPoolEntry(PoolBase.java:206)
\tat com.zaxxer.hikari.pool.HikariPool.createPoolEntry(HikariPool.java:476)
\tat com.zaxxer.hikari.pool.HikariPool.checkFailFast(HikariPool.java:561)
\tat com.zaxxer.hikari.pool.HikariPool.<init>(HikariPool.java:115)
\tat com.zaxxer.hikari.HikariDataSource.getConnection(HikariDataSource.java:112)
{code}
The solution can be exposing the apache drill pod host name to kubernetes
cluster. But is there anything we are missing or improvement needed, as a part
of zookeeper and apache drill configuration setup.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
