AnsweredAssumed Answered

Using JDBC with SQL Workbench/J for connecting to Drill

Question asked by Arunav on Mar 27, 2017
Latest reply on Mar 28, 2017 by Arunav

Hi,

I was working on the configuration of SQL Workbench/J.  The SQL Workbench/J is installed on a Windows 2008 R2 server(On Aws). I've a 3 node MapR cluster(On Aws). 3 Zookeepers running and 2 Drillbits. Here's the details from the drill-override.conf:

drill.exec: {
cluster-id: "mapr_cluster_com-drillbits",
zk.connect: "ip-172-31-51-184.ec2.internal:5181,ip-172-31-52-143.ec2.internal:5181,ip-172-31-60-214.ec2.internal:5181"
}

 

On the Windows m/c, I've downloaded Drill JDBC41 Driver Jar files from Downloading and Configuring the Driver. And I've added the drivers on the SQL Workbench/J. Document followed for this is Using the JDBC Driver - Apache Drill 

 

Now, when I provide the URL for a single Zookeeper node, I get the following error.
Note: The server name provided in the URL has an active drillbit running on it.

 

But, with the same JDBC driver, when I provide all 3 zookeeper nodes in the URL, my connection is successful.

jdbc:drill:zk=ip-172-31-51-184.ec2.internal:5181,ip-172-31-52-143.ec2.internal:5181,ip-172-31-60-214.ec2.internal:5181/drill/mapr_cluster_com-drillbits;schema=hive

 

So, my questions are:
1. Why the connection is not successful with a single zookeeper, but it's successful when all 3 zookeepers are provided?
2. Though the  connection to drill is getting established, there's a significant time lag of 5-7 mins before it's getting      established. Why this delay? I have configured and tried this process on AWS as well as on physical servers(No firewall or proxy configurations are there).
Note: Once the connection is established, there's no delay in running queries on databases(Tested on hive DBs).

 

 

Please help me understand the architecture better.

 

Thanks,

Arunava Addy

Outcomes