topic Connect to Databricks using Java SDK through proxy in Get Started Discussions

Connect to Databricks using Java SDK through proxy

Nagasundaram — Mon, 12 Feb 2024 14:05:33 GMT

I'm trying to connect to databricks from java using the java sdk and get cluster/sqlWarehouse state. I'm able to connect and get cluster state from my local. But, once I deploy it to the server, my company's network is not allowing the connection. We need to use proxy here but I'm not sure how to use proxy with the databricks java sdk.

Below is the code that works in local env:

DatabricksConfig config = new DatabricksConfig().setHost("https://name.databricks.com").setToken("myToken").resolve();

WorkspaceClient wc = new WorkspaceClient (config); wc.clusters().get("myClusterId").getState().toString();

Any hint or suggestion would be very helpful.

Re: Connect to Databricks using Java SDK through proxy

Allia — Wed, 14 Feb 2024 10:13:10 GMT

Hi @Nagasundaram

You can make use of the below init script inorder to use a proxy server with Databricks cluster. The content of the init script can be added at "Workspace/shared/setproxy.sh"

==================================================

val proxy = "http://localhost:8888" // set this to your actual proxy

val proxy_host = "localhost"

val proxy_port = "8888"

val no_proxy = "127.0.0.1,.local,169.254.169.254,s3.amazonaws.com,s3.us-east-1.amazonaws.com" // make sure to update no proxy as needed (e.g. for S3 region or any other internal domains)

dbutils.fs.put("Workspace/shared/setproxy.sh", s"""#!/bin/bash

echo "export http_proxy=$proxy" >> /databricks/spark/conf/spark-env.sh

echo "export https_proxy=$proxy" >> /databricks/spark/conf/spark-env.sh

echo "export no_proxy=$no_proxy" >> /databricks/spark/conf/spark-env.sh

echo "export HTTP_PROXY=$proxy" >> /databricks/spark/conf/spark-env.sh

echo "export HTTPS_PROXY=$proxy" >> /databricks/spark/conf/spark-env.sh

echo "export NO_PROXY=$no_proxy" >> /databricks/spark/conf/spark-env.sh

echo "export _JAVA_OPTIONS=\"-Dhttps.proxyHost=${proxy_host} -Dhttps.proxyPort=${proxy_port} -Dhttp.proxyHost=${proxy_host} -Dhttp.proxyPort=${proxy_port} -Dhttp.nonProxyHosts=${java_no_proxy}\"" >> /databricks/spark/conf/spark-env.sh

echo "http_proxy=$proxy" >> /etc/environment

echo "https_proxy=$proxy" >> /etc/environment

echo "no_proxy=$no_proxy" >> /etc/environment

echo "HTTP_PROXY=$proxy" >> /etc/environment

echo "HTTPS_PROXY=$proxy" >> /etc/environment

echo "NO_PROXY=$no_proxy" >> /etc/environment

cat >> /etc/R/Renviron << EOF

http_proxy=$proxy

https_proxy=$proxy

no_proxy=$no_proxy

EOF

""", true)

==================================================

Please test it carefully in your environment.

Let me know if that helps.

Re: Connect to Databricks using Java SDK through proxy

Nagasundaram — Fri, 16 Feb 2024 11:58:27 GMT

Hi @Retired_mod

Thanks for the reply. I tried adding the arguments in manifest.yml file like this:

JAVA_OPTS: "-Dhttp.proxyHost='your_proxy_hos't -Dhttp.proxyPort='your_proxy_port' -Dhttps.proxyHost='your_proxy_host' -Dhttps.proxyPort='your_proxy_port'"

Still, I'm getting connection refused to databricks when I deploy this to PCF.

Is using WorkspaceClient the right way? Or, do I need to use AccountClient?