SAS Data Integration Studio, DataFlux Data Management Studio, SAS/ACCESS, SAS Data Loader for Hadoop and others

Hadoop schema not found

Accepted Solution Solved
Reply
Frequent Contributor
Posts: 93
Accepted Solution

Hadoop schema not found

[ Edited ]

Hi team,

 

Using SAS Access to Hadoop on RHEL 6 running SAS 9.4 M3 connecting to Hive on HDP.

 

Since this site uses ZooKeeper as services discovery, I'm using the URI option in the libname statement. The URI string has been validated outside of SAS and working. I can access Ambari and run queries against the Hive table with no issues.

 

However, when connecting from SAS, I keep on getting 'schema not found' when trying to assign the libname.

Doesn't matter what I do, I keep getting this.

 

Any ideas ?

 

Thanks,


Accepted Solutions
Solution
Thursday
SAS Employee
Posts: 203

Re: Hadoop schema not found

Hi @nhvdwalt

 

I think that I can explain this. Keberos user name may not be case sensitive. This could allow you to connect.  The Hive schemas are tied to HDFS file system names and are case sensitive. So "BOB" does not equal "bob". The problem is when you try to access your data the schema names (really HDFS file system) do not match because they are not the same case. This results in a "schema not found" message.

 

Best wishes,

Jeff

View solution in original post


All Replies
SAS Employee
Posts: 203

Re: Hadoop schema not found

Hi @nhvdwalt

 

By "services discovery" do you mean Hive High Availability?

 

If so, this is not supported at SAS 9.4M3. It is a new feature for SAS 9.4M4. The connection must be made to a specific Hive at M3.

 

Best wishes,

Jeff

Frequent Contributor
Posts: 93

Re: Hadoop schema not found

Hi Jeff,

 

Thanks, I actually forgot to post the solution.

 

I'm not a Hadoop expert....so this is my understanding. At our site we cannot connect directly to Hive. We connect to Hive via ZooKeeper, hence the need for the URI option. The HA topic has never come up, so I would assume it's not that.

 

The cause of the problem, was the credentials I connect to Hadoop with. This site uses Kerberos. If I generate the TGT with my user in upper case, all is ok and I can see the Hive database. If I generate my TGT in lowercase, it fails with the above error. It's bizarre. For some reason Hadoop only authorises my when I'm uppercase, but will authenticate me either way. Funny enough, Ambari doesn't care how I sign on and I can see all resources I have been granted access to.

Solution
Thursday
SAS Employee
Posts: 203

Re: Hadoop schema not found

Hi @nhvdwalt

 

I think that I can explain this. Keberos user name may not be case sensitive. This could allow you to connect.  The Hive schemas are tied to HDFS file system names and are case sensitive. So "BOB" does not equal "bob". The problem is when you try to access your data the schema names (really HDFS file system) do not match because they are not the same case. This results in a "schema not found" message.

 

Best wishes,

Jeff

Frequent Contributor
Posts: 93

Re: Hadoop schema not found

Great, thanks @JBailey

SAS Employee
Posts: 203

Re: Hadoop schema not found

My pleasure, @nhvdwalt

☑ This topic is SOLVED.

Need further help from the community? Please ask a new question.

Discussion stats
  • 5 replies
  • 200 views
  • 0 likes
  • 2 in conversation