question Re: Impala queries through impala JDBC is running with empty user in Support Questions

Impala queries through impala JDBC is running with empty user

Fawze — Fri, 16 Sep 2022 10:57:39 GMT

Hi All,

I'm using zeppelin to run report using impala.

I downloaded impala JDBC to my linux machine, when i'm looking at the impala queries at cloudera manager, i see my queries running with empty user " ".

I didn't integrate LDAP authnitication yet with Impala.

I have 2 small questions:

1- where and how i can pass the user that i need the queries to run with in the impala JDBC?

2- How i can restrict only queries qith user specific to run.

Re: Impala queries through impala JDBC is running with empty user

saranvisa — Thu, 26 Jan 2017 02:48:00 GMT

@Fawze

The answer for your 2nd question first, You have to implement apache sentry to restrict user specific query access on impala. Work with your Hadoop admin to setup it on your cluster

now for the first (generic) question, obviously you have to pass the parameter from the source. Apache sentry has dependency on other security tool called kerberos which has a concept called keytab. so when you pass the user and password from the source, you also have to pass the keytab to authenticate your network. you can achieve your first requirement with this step

I've shared some high level security information in the below link, hope this will give some idea

https://community.cloudera.com/t5/Security-Apache-Sentry/Hadoop-Security-for-beginners/m-p/49876#M247

Thanks

Kumar

Re: Impala queries through impala JDBC is running with empty user

Fawze — Thu, 26 Jan 2017 03:04:14 GMT

I'm already working to integrate authentication and security to my cluster,
it may takes time for me to complete this process, but my question where i
can provide user name in the impala JDBC if it in a linux user.

Re: Impala queries through impala JDBC is running with empty user

saranvisa — Thu, 26 Jan 2017 03:30:43 GMT

@Fawze

I think you didn't get my point yet,

You don't need to provide user name for impala instead you can authenticate your table in two steps
1. Network authentication using Kerberos
2. Query/Table level authentication using Apache Sentry

so when you call the query using JDBC, you need to pass user, passworld & keytab

Re: Impala queries through impala JDBC is running with empty user

Fawze — Thu, 26 Jan 2017 03:38:35 GMT

Hi saranvisa,

I got your point, but currently i don't have any authentication that i'm using.

How i can enforce the queries running through specific impala JDBC to run with specific user.

Maybe i'm missing something in the concept.

Re: Impala queries through impala JDBC is running with empty user

Fawze — Thu, 26 Jan 2017 19:46:12 GMT

Anyone who can help with this please.

Re: Impala queries through impala JDBC is running with empty user

Fawze — Tue, 31 Jan 2017 02:21:07 GMT

This is my default.url and impala.url in the interpreter settings:

jdbc:impala://xxxxxx:21050/;SID=fawze;

Re: Impala queries through impala JDBC is running with empty user

Lars Volker — Wed, 01 Feb 2017 00:37:14 GMT

Hi Fawze,

which JDBC driver are you using? And which version of it?

Cheers, Lars

Re: Impala queries through impala JDBC is running with empty user

Fawze — Wed, 01 Feb 2017 04:25:53 GMT

ImpalaJDBC4.jar

Re: Impala queries through impala JDBC is running with empty user

Lars Volker — Wed, 01 Feb 2017 16:03:21 GMT

That version should have been downloaded from Cloudera and as such the downloaded file comes with its own version number. The current version is 2.5.36 and can be found here: http://www.cloudera.com/downloads/connectors/impala/jdbc/2-5-36.html

Please have a look at the documentation contained in that download ("Cloudera-JDBC-Driver-for-Impala-Install-Guide.pdf"), especially at the section "Using User Name" and the "UID" parameter.

Cheers, Lars

Re: Impala queries through impala JDBC is running with empty user

Fawze — Sat, 04 Feb 2017 04:04:28 GMT

Already tried this, I don't know how some can run query without user.

When i'm looking at CM i see akot of queries running with empty user and cann't localize the queries source and user, which make me hesitate to stop impala at my production cluster.

Re: Impala queries through impala JDBC is running with empty user

Fawze — Mon, 06 Feb 2017 20:51:15 GMT

Anyone here who is familiar with apache zeppelin?

Re: Impala queries through impala JDBC is running with empty user

Fawze — Sat, 18 Feb 2017 07:46:20 GMT

Any help?

Re: Impala queries through impala JDBC is running with empty user

Lars Volker — Tue, 21 Feb 2017 18:33:48 GMT

Hi Fawze,

what version of the Impala JDBC driver are you using? What did you already try? The documentation of the latest version of the JDBC driver does not mention a "SID" parameter, but your connection string does. However, the documentation describes a "UID" parameter. Can you try using that one instead?

Cheers, Lars

Re: Impala queries through impala JDBC is running with empty user

Fawze — Wed, 22 Feb 2017 02:23:38 GMT

Hi Lars,

In your comments before you asked me to check the UID and i mentioned the version 'm using, unfortunately, it doesn't work.

Re: Impala queries through impala JDBC is running with empty user

Lars Volker — Tue, 28 Feb 2017 21:47:55 GMT

Hi Fawze,

That version should have been downloaded from Cloudera and as such the downloaded file comes with its own version number. The current version is 2.5.36. "ImpalaJDBC4.jar" is the name of the file, not the version.

What is the connection string you used? What was the error you received?

Thanks, Lars

Re: Impala queries through impala JDBC is running with empty user

Fawze — Tue, 28 Feb 2017 22:04:16 GMT

jdbc:impala://ropr-dhc001:21050/;UID=fawze

Cloudera_ImpalaJDBC_2.5.5.1007.zip

No erros but the queries are running without user.

Re: Impala queries through impala JDBC is running with empty user

Lars Volker — Tue, 28 Feb 2017 22:35:18 GMT

Your version of the driver seems to be 2.5.5. The current version is 2.5.36 and can be found here: http://www.cloudera.com/downloads/connectors/impala/jdbc/2-5-36.html

Please try again with the latest version and see if that help. The UID parameter should be supported there.

Re: Impala queries through impala JDBC is running with empty user

mihirkk — Wed, 15 Nov 2017 01:28:42 GMT

Using Cloudera JDBC41_2.5.38.

Tried various combinations but no luck using JDBC, user is empty.

DataSource ds = new com.cloudera.impala.jdbc41.DataSource();
ds.setURL("jdbc:impala://<host>:21050;UseSasl=0;UID=test;AuthMech=0");
ds.setUserID("test");
ds.setLoginTimeout(60);
ds.setCustomProperty("UID", "test");
ds.setCustomProperty("USER", "test");
ds.setLogLevel(LogLevel.TRACE.toString());
ds.setLogDirectory("C:\\temp\\impala_log");
String sql = "select ....";

    try (Connection connection = ds.getConnection(); ResultSet rs = connection.createStatement().executeQuery(sql) {

    // print 

We do not have authentication turned on so I do get result back but "user" is empty on impalad host http://<impaladhost>:25000/queries
25000/queries



I see that in python -> impala/hiveserver2.py it tries to read from os environment variable ('USER', 'USERNAME') etc.
I tried to do TRACE log for cloudera but could not make much sense of logs.

If I use hue from cloudera manager then I do see user in impalad.

Any idea what can I do in Java JDBC without turning on User or User/Password authentication.

Re: Impala queries through impala JDBC is running with empty user

trock — Fri, 19 Jan 2018 01:45:19 GMT

I have exactly the same empty username issue. My driver is Impala JDBC v2.5.41 and I am not using username/password authentication. Not having the user name makes searching queries by user impossible. Anyone here can help?