Not able to install hadoop using Cloudera Manager

Question

I am trying to setup hadoop cluster in a single VM (for simplicity) using cloudera Manager 5.9. The below are the details of my environment:

Host OS -> Windows 10
Virtualization software -> Virtual box 5.1.10
Guest OS -> Cent OS 6.8

I installed the Cloudera Manager, and as per instruction of Cloudera Manager did follow the steps.

Most of the installation steps went fine, but at the last check it fails.The below is the screen shot of that.

As can be seen in screen shot, it is giving error:
"Unexpected error. Unable to verify database connection."

I have made the necessary changes in the configuration files of Postgres DB which Cloudera uses by default, that is it should be able to accept the remote connections.

There are no errors in the logs of Cloudera manager. I did online search as well but no success.

Can anyone help me solving this error?

Answer 1

First of all, make sure you started the embedded DB. As the other tests are skipped, maybe your server is down. :

sudo service cloudera-scm-server-db start

I don't think it's your case, as CM seems to be working.

So, if you already started your embedded database, and still have these kind of issues, problems usually came from 2 different sides :

• DB problems: It can be obvious, but I don't think it's your case. In my case, I didn't configure pg_hba.conf to make my postgresql be accessed by the incriminated role (here, RMAN). But as I said, you seem to be using the embedded cloudera postgres, so I don't think it's your case.
• Python issue: Sometimes, due to a reason I can't control, some symlinks to python libraries are lost during the install process. When you already install your cluster, hue is a great candidate for being impacted, but Cloudera's services are good candidates too, when you're trying to configure your DB. In this case, check your cloudera's agent log (/var/log/cloudera-scm-agent/cloudera-scm-agent.log) and you should see something about psycopg2. If it's the case, you only have to create the following symlink:

sudo ln -s /usr/lib64/python2.7/site-packages/psycopg2 /opt/cloudera/parcels/CDH/lib/hue/build/env/lib/python2.7/site-packages/psycopg2

Anyway, If you're trying to understand how to install CDH, I recommend you to not use the embedded Postgres DB, but Postgres as external DB instead(in the same machine). Cloudera Manager is using some scripts to generate lots of things (passwords, configs, ...) you're going to miss if you use the embedded version of postgres.

On the other hand, if you're only want to access a single-node CDH 5.9+ clúster to make some tests, download the quickstart and follow the cloudera guide to upgrade the cluster to a newer version.