Is it possible to disable jdbc/odbc connection to (azure) databrick cluster

[OvZ]

Hi,

I wanna know if it is possible to disable jdbc/odbc connection to (azure) databrick cluster.

So know (download) tools could connect this way ?

Thz in adv,

Oscar

[LandanG · Accepted solution]

@Oscar van Zelst​ apart from using configuration settings to restrict access, you can also use an init script to disable connections on a cluster. The script is:

%scala
val initScriptContent = s"""
 |#!/bin/bash
 |cat > /databricks/common/conf/disable_jdbc_odbc.conf << EOL
 |{
 |  databricks.daemon.driver.startJDBC = false
 |}
 |EOL
""".stripMargin
dbutils.fs.put("dbfs:/databricks/init_scripts/disable_jdbc_odbc.conf",initScriptContent, true)

[OvZ]

Thank you for you reply this sounds great but I get a error (sorry I am a newbie) or should I put this as a global init script ?

[LandanG]

Hi @Oscar van Zelst​ ,

You can use Cluster-scoped init scripts in the cluster configuration as seen in the screenshot or these docs.

 

[OvZ]

The error I get if I put this as global inti script is that the cluster won't start

[LandanG]

Can you try using cluster-scoped init scripts?

[OvZ]

I configure this now (see below) but it give a error

[LandanG]

@Oscar van Zelst​ Try using a cluster-scoped init script

[OvZ]

Hi,

I have following you advise but no luck. This steps I have made; On cluster level I have done this ; Even in de web terminal I see some files; root@0123-123553-quwe076y-10-215-1-4:/databricks/init_scripts# ls -ltr

total 4

-rw-r--r-- 1 root root 43 Jan 31 07:43 20230131_074358_00_disable_jdbc_odbc.sh.stdout.log

-rw-r--r-- 1 root root 0 Jan 31 07:43 20230131_074358_00_disable_jdbc_odbc.sh.stderr.log

root@0123-123553-quwe076y-10-215-1-4:/databricks/init_scripts# more 20230131_074358_00_disable_jdbc_odbc.sh.stdout.log

databricks.daemon.driver.startJDBC = false

But still I am able to connect by jdbc what I ‘am doing wrong ?

[OvZ]

even if is set it on this way it does nothing.. way

[OvZ]

Also in /tmp/custom-spark.conf you can see its set ;

root@0126-125139-sxrs37zl-10-215-1-5:/databricks/driver# cat /tmp/custom-spark.conf

...

spark.master local[*, 4]

spark.databricks.passthrough.enabled true

spark.databricks.cloudfetch.hasRegionSupport true

spark.databricks.daemon.driver.startJDBC false

[LandanG]

@Oscar van Zelst​ Are you saving the code as a .conf file, as seen in the code snippet?

dbutils.fs.put("dbfs:/databricks/init_scripts/disable_jdbc_odbc.conf",initScriptContent, true)

[OvZ]

Hi ,

I have done this and this and if I look at terminal level its there;

root@0123-123553-quwe076y-10-215-1-4:/databricks/init_scripts# ls -ltr

total 4

-rw-r--r-- 1 root root 0 Jan 31 16:04 20230131_160437_00_disable_jdbc_odbc.conf.stdout.log

-rw-r--r-- 1 root root 0 Jan 31 16:04 20230131_160437_00_disable_jdbc_odbc.conf.stderr.log

-rw-r--r-- 1 root root 49 Jan 31 16:04 disable_jdbc_odbc.conf

root@0123-123553-quwe076y-10-215-1-4:/databricks/init_scripts# cat disable_jdbc_odbc.conf 

{

 databricks.daemon.driver.startJDBC = false

}

but still accesable by jdbc/odbc .connection it there other methodes to disable access also fine

[wdphilli]

Hi @OvZ & @LandanG, as a point of clarification, the script provided is intended to be run in a notebook first. After running the below in a notebook, it creates the init script at the location "dbfs:/databricks/init_scripts/disable_jdbc_odbc.conf"

%scala
val initScriptContent = s"""
 |#!/bin/bash
 |cat > /databricks/common/conf/disable_jdbc_odbc.conf << EOL
 |{
 |  databricks.daemon.driver.startJDBC = false
 |}
 |EOL
""".stripMargin
dbutils.fs.put("dbfs:/databricks/init_scripts/disable_jdbc_odbc.conf",initScriptContent, true)

 From there, it can be added as a cluster-scoped init script under Advanced options:

After cluster startup, you will notice that the JDBC/ODBC tab does not appear in the Spark UI:

JDBC/ODBC connections will also fail with a 502 error: